M.Sc. Courses at TU Delft
(AE4317) Autonomous Flight of Micro Air Vehicles
This course covers the challenges and existing state-of-the-art methods for enabling autonomous flight of Micro Air Vehicles (MAVs), ranging from 20-gram flapping wings to 1 kg quad rotors. The emphasis is on computationally efficient, bio-inspired approaches to MAV autonomous flight. The theoretical knowledge will be applied by the students in the practical assignment, in which student groups program quad rotors in order to avoid obstacles in TU Delft’s Cyberzoo.
https://www.youtube.com/watch?v=R3LNWh3UQHk
(AE4350) Bio-Inspired Intelligence and Learning for Autonomous Applications
This course covers the following subjects:
- Introduction Bio-inspired Intelligence
- Introduction to Artificial Neural Networks
- Introduction to Reinforcement Learning
- Reinforcement Learning for Aerospace Control
- Evolutionary robotics
- Self-supervised learning
M.Sc. Thesis at TU Delft
The MAVLab guides Master students from several faculties during their Master’s thesis. Please get in contact with us if you would like to pursue a thesis project with us.
Have a look at our recent list of M.Sc. thesis subjects:
2017
Grebe, Nicolás Omar Abuter
Differential Dynamic Programming for Aerial Robots using an Aerodynamics Model Masters Thesis
TU Delft Aerospace Engineering; TU Delft Control & Simulation, 2017, (de Wagter, C. (mentor); Delft University of Technology (degree granting institution)).
@mastersthesis{uuid:edbb8630-d1ad-4230-b4cd-f593e81622b2,
title = {Differential Dynamic Programming for Aerial Robots using an Aerodynamics Model},
author = {Nicolás Omar Abuter Grebe},
url = {http://resolver.tudelft.nl/uuid:edbb8630-d1ad-4230-b4cd-f593e81622b2},
year = {2017},
date = {2017-01-01},
school = {TU Delft Aerospace Engineering; TU Delft Control & Simulation},
abstract = {State of the art trajectory generation schemes for quadrotors assume a simple dynamic model. They neglect aerodynamic effects such as induced drag and blade flapping and assume that no wind is present. In order to overcome this limitation, this thesis investigates a trajectory optimization scheme based upon Differential Dynamic Programming (DDP). There are various software-implementations of the DDP scheme. For future deployment on robotic hardware the software is required to be computationally efficient, written in C++ and to be open-source. A library named GCOP, which was developed at the John Hopkins University, fulfills these requirements and is used. Before implementing the solver, a full model of the Crazyflie Nano Quadcopter is identified experimentally. The solver is validated, normalized and the performance is benchmarked. This method yields reliable minimum control-effort trajectories. A control scheme is proposed and studied in Monte-Carlo simulations. Itis robust and able to handle large modelling errors in mass and moment of inertia while ensuring minimal error on the final state.},
note = {de Wagter, C. (mentor); Delft University of Technology (degree granting institution)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
State of the art trajectory generation schemes for quadrotors assume a simple dynamic model. They neglect aerodynamic effects such as induced drag and blade flapping and assume that no wind is present. In order to overcome this limitation, this thesis investigates a trajectory optimization scheme based upon Differential Dynamic Programming (DDP). There are various software-implementations of the DDP scheme. For future deployment on robotic hardware the software is required to be computationally efficient, written in C++ and to be open-source. A library named GCOP, which was developed at the John Hopkins University, fulfills these requirements and is used. Before implementing the solver, a full model of the Crazyflie Nano Quadcopter is identified experimentally. The solver is validated, normalized and the performance is benchmarked. This method yields reliable minimum control-effort trajectories. A control scheme is proposed and studied in Monte-Carlo simulations. Itis robust and able to handle large modelling errors in mass and moment of inertia while ensuring minimal error on the final state.
Rijks, F. G. J.
Studying the effect of the tail on the dynamics of a flapping-wing MAV Masters Thesis
Delft University of Technology, 2017, (De Visser, C.C. (mentor); Karásek, M. (mentor); Armanini, S.F. (mentor)).
@mastersthesis{uuid:18dee61c-9828-430a-9d71-5a12586da89c,
title = {Studying the effect of the tail on the dynamics of a flapping-wing MAV},
author = {F. G. J. Rijks},
url = {http://resolver.tudelft.nl/uuid:18dee61c-9828-430a-9d71-5a12586da89c},
year = {2017},
date = {2017-01-01},
school = {Delft University of Technology},
abstract = {The effects of horizontal tail geometry and position on longitudinal flapping-wing micro aerial vehicle dynamics were studied using wind tunnel and free-flight experiments. Linearised models were used to analyse the effect on the dynamic properties of the ornithopter. Results show higher steady-state velocity and increased pitch damping for increased tail surface area and aspect ratio. The maximum span width of the tail surface is also found to play an important role in determining dynamic behaviour, in particular when the distance between the tail surface and the flapping wings is large. Steady-state conditions can be predicted accurately using linear functions of tail geometry for this ornithopter. Predicting dynamic behaviour is more complicated and requires further study. However, the observed trends in some of the model parameters suggest that future models explicitly including the tail geometry may be used to design flapping-wing robots with desirable dynamic properties.},
note = {De Visser, C.C. (mentor); Karásek, M. (mentor); Armanini, S.F. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
The effects of horizontal tail geometry and position on longitudinal flapping-wing micro aerial vehicle dynamics were studied using wind tunnel and free-flight experiments. Linearised models were used to analyse the effect on the dynamic properties of the ornithopter. Results show higher steady-state velocity and increased pitch damping for increased tail surface area and aspect ratio. The maximum span width of the tail surface is also found to play an important role in determining dynamic behaviour, in particular when the distance between the tail surface and the flapping wings is large. Steady-state conditions can be predicted accurately using linear functions of tail geometry for this ornithopter. Predicting dynamic behaviour is more complicated and requires further study. However, the observed trends in some of the model parameters suggest that future models explicitly including the tail geometry may be used to design flapping-wing robots with desirable dynamic properties.
2016
Hordijk, B J Pijnacker
Vertical Landing for Micro Air Vehicles using Event-Based Optical Flow Masters Thesis
Delft University of Technology, Delft, NL, 2016.
@mastersthesis{Pijnacker2016,
title = {Vertical Landing for Micro Air Vehicles using Event-Based Optical Flow},
author = {B J Pijnacker Hordijk},
url = {http://resolver.tudelft.nl/uuid:ffa1ec41-3930-4dfe-b454-e11c3517a7f4},
year = {2016},
date = {2016-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
der Sman, E. S. Van
Delft University of Technology, 2016, (Chu, Q.P. (mentor); Remes, B. (mentor); Smeur, E.J.J. (mentor)).
@mastersthesis{uuid:b76bd35d-9d56-472e-8ff8-35fd453b6a49,
title = {Incremental Nonlinear Dynamic Inversion and Multihole Pressure Probes for Disturbance Rejection Control of Fixed-Wing Micro Air Vehicles},
author = {E. S. Van der Sman},
url = {http://resolver.tudelft.nl/uuid:b76bd35d-9d56-472e-8ff8-35fd453b6a49},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {Maintaining stable flight during high turbulence intensities is challenging for fixed-wing micro air vehicles. Two methods have been identified to improve the disturbance rejection performance of the MAV: incremental nonlinear dynamic inversion and phase-advanced pitch probes. Incremental nonlinear dynamic inversion uses the angular acceleration measurements to counteract disturbances. Multihole pressure probes measure the incoming flow angle and velocity ahead of the wing in order to react to gusts before an inertial response has occurred. The performance of incremental nonlinear dynamic inversion is compared to a traditional proportional integral derivative controller with and without the multihole pressure probes. The attitude controllers are tested by performing autonomous wind tunnel flights and stability augmented outdoor flights. This thesis shows that nonlinear dynamic inversion improves the disturbance rejection performance of fixed-wing MAVs compared to traditional proportional integral derivative controllers.},
note = {Chu, Q.P. (mentor); Remes, B. (mentor); Smeur, E.J.J. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Maintaining stable flight during high turbulence intensities is challenging for fixed-wing micro air vehicles. Two methods have been identified to improve the disturbance rejection performance of the MAV: incremental nonlinear dynamic inversion and phase-advanced pitch probes. Incremental nonlinear dynamic inversion uses the angular acceleration measurements to counteract disturbances. Multihole pressure probes measure the incoming flow angle and velocity ahead of the wing in order to react to gusts before an inertial response has occurred. The performance of incremental nonlinear dynamic inversion is compared to a traditional proportional integral derivative controller with and without the multihole pressure probes. The attitude controllers are tested by performing autonomous wind tunnel flights and stability augmented outdoor flights. This thesis shows that nonlinear dynamic inversion improves the disturbance rejection performance of fixed-wing MAVs compared to traditional proportional integral derivative controllers.
Janssen, Y S
Reinforcement Learning Policy Approximation by Behavior Trees: using Genetic Algoritms Masters Thesis
Delft University of Technology, Delft, NL, 2016.
@mastersthesis{Janssen2016,
title = {Reinforcement Learning Policy Approximation by Behavior Trees: using Genetic Algoritms},
author = {Y S Janssen},
url = {http://resolver.tudelft.nl/uuid:f6008da9-d688-4b9f-9880-8d7c3b51a777},
year = {2016},
date = {2016-01-01},
urldate = {2016-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Fonville, C R
Delft University of Technology, Delft, NL, 2016.
@mastersthesis{Fonville2016,
title = {The Exploring DelFly: How to increase the indoor explored area of the DelFly Explorer by means of computationally efficient routing decisions?},
author = {C R Fonville},
url = {http://resolver.tudelft.nl/uuid:8efab9c5-e78b-40ff-ab37-a563366d22f9},
year = {2016},
date = {2016-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Kuijpers, M. W. M.
The influence of a bottom camera in indoor ground-segmentation based obstacle avoiding performance for MAVs Masters Thesis
Delft University of Technology, 2016, (de Wagter, C. (mentor); de Croon, G.C.H.E. (mentor)).
@mastersthesis{uuid:424ead9b-50be-4e80-94a9-d041a1418dd3,
title = {The influence of a bottom camera in indoor ground-segmentation based obstacle avoiding performance for MAVs},
author = {M. W. M. Kuijpers},
url = {http://resolver.tudelft.nl/uuid:424ead9b-50be-4e80-94a9-d041a1418dd3},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
note = {de Wagter, C. (mentor); de Croon, G.C.H.E. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Lamers, K
Self-Supervised Monocular Distance Learning on a Lightweight Micro Air Vehicle Masters Thesis
Delft University of Technology, Delft, NL, 2016.
@mastersthesis{Lamers2016b,
title = {Self-Supervised Monocular Distance Learning on a Lightweight Micro Air Vehicle},
author = {K Lamers},
url = {http://resolver.tudelft.nl/uuid:55f9ab7a-2651-4a90-93a0-a8c9ddc7c6a9},
year = {2016},
date = {2016-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Höppener, D. C.
Actuator Saturation Handling using Weighted Optimal Control Allocation Applied to an INDI Controlled Quadcopter Masters Thesis
Delft University of Technology, 2016, (de Wagter, C. (mentor)).
@mastersthesis{uuid:3704b044-b9bf-454a-8678-0d140bd1d308,
title = {Actuator Saturation Handling using Weighted Optimal Control Allocation Applied to an INDI Controlled Quadcopter},
author = {D. C. Höppener},
url = {http://resolver.tudelft.nl/uuid:3704b044-b9bf-454a-8678-0d140bd1d308},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {Incremental Nonlinear Dynamic Inversion provides a high performance attitude controller for multi-rotor Micro Aerial Vehicles by providing very good disturbance rejection capabilities. Flights conducted with a quadcopter revealed undesired pitch and rolling motions which occurred simultaneously with actuator saturation for instantaneous yaw angle reference tracking commands. Constrained control allocation methods can increase the system's performance by providing an effective strategy to prioritize control objectives, and redistribute control effort accordingly. Weighted Least Squares control allocation makes the constrained control allocation problem a quadratic optimization problem. An iterative solver based on the computationally efficient active-set algorithm finds the optimal control distribution for a weighted control objective. In this paper the Weighted Least Squares control allocator is used to overcome two challenges 1) increase performance by applying prioritization between control objectives and redistribute control effort accordingly, accounting for the actuator limits 2) enable flight when flying with severely compromised actuator(s). Real-world flight experiments are performed and show a significant increase in performance for high load yaw maneuvers, and enabled a quadcopter to perform controlled flight with a severely compromised actuator},
note = {de Wagter, C. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Incremental Nonlinear Dynamic Inversion provides a high performance attitude controller for multi-rotor Micro Aerial Vehicles by providing very good disturbance rejection capabilities. Flights conducted with a quadcopter revealed undesired pitch and rolling motions which occurred simultaneously with actuator saturation for instantaneous yaw angle reference tracking commands. Constrained control allocation methods can increase the system's performance by providing an effective strategy to prioritize control objectives, and redistribute control effort accordingly. Weighted Least Squares control allocation makes the constrained control allocation problem a quadratic optimization problem. An iterative solver based on the computationally efficient active-set algorithm finds the optimal control distribution for a weighted control objective. In this paper the Weighted Least Squares control allocator is used to overcome two challenges 1) increase performance by applying prioritization between control objectives and redistribute control effort accordingly, accounting for the actuator limits 2) enable flight when flying with severely compromised actuator(s). Real-world flight experiments are performed and show a significant increase in performance for high load yaw maneuvers, and enabled a quadcopter to perform controlled flight with a severely compromised actuator
Janssen, R. M. J.
Attitude control- and stabilisation moment generation of the DelFly using Wing Tension Modulation Masters Thesis
Delft University of Technology, 2016, (Karasek, M. (mentor)).
@mastersthesis{uuid:382dec56-7789-40df-af28-f2e61de99fad,
title = {Attitude control- and stabilisation moment generation of the DelFly using Wing Tension Modulation},
author = {R. M. J. Janssen},
url = {http://resolver.tudelft.nl/uuid:382dec56-7789-40df-af28-f2e61de99fad},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
note = {Karasek, M. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Duro, T.
Tracking and Following a Moving Person Onboard a Small Pocket Drone Masters Thesis
Delft University of Technology, 2016, (De Croon, G. (mentor); De Wagter, C (mentor); Meertens, R. (mentor)).
@mastersthesis{uuid:58a4c285-e3b6-4bf0-b885-2908077e9b02,
title = {Tracking and Following a Moving Person Onboard a Small Pocket Drone},
author = {T. Duro},
url = {http://resolver.tudelft.nl/uuid:58a4c285-e3b6-4bf0-b885-2908077e9b02},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {This paper presents a vision based strategy, designed to work fully onboard a small pocket drone, for autonomously tracking and following a person. Flying a drone is not an easy task, usually requiring a trained pilot, with the presented system it is possible to use a drone for filming or taking pictures from previously inaccessible places without the need for a person controlling the aircraft. Such framework is comprised by two main components, a tracker and a control system. The tracker has the function of estimating the position of the person that is being followed, while the control system gets the drone near that person. Limited by payload weight, power consumption and processing power the system results in a delicate balance between these constraints. The main contributions of this paper are the comparison between two state-of-the-art visual trackers running on paparazzi, Struck and KCF, as well as the control system that uses the tracker’s output location to perform the person following task. Then a new tracker is developed to be as computationally light as possible so that it can run onboard a small pocket drone, based on HOG feature extraction, it uses logistic regression to train a detector on the appearance of a person.},
note = {De Croon, G. (mentor); De Wagter, C (mentor); Meertens, R. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
This paper presents a vision based strategy, designed to work fully onboard a small pocket drone, for autonomously tracking and following a person. Flying a drone is not an easy task, usually requiring a trained pilot, with the presented system it is possible to use a drone for filming or taking pictures from previously inaccessible places without the need for a person controlling the aircraft. Such framework is comprised by two main components, a tracker and a control system. The tracker has the function of estimating the position of the person that is being followed, while the control system gets the drone near that person. Limited by payload weight, power consumption and processing power the system results in a delicate balance between these constraints. The main contributions of this paper are the comparison between two state-of-the-art visual trackers running on paparazzi, Struck and KCF, as well as the control system that uses the tracker’s output location to perform the person following task. Then a new tracker is developed to be as computationally light as possible so that it can run onboard a small pocket drone, based on HOG feature extraction, it uses logistic regression to train a detector on the appearance of a person.
Paquim, J.
Learning Depth from Single Monocular Images Using Stereo Supervisory Input Masters Thesis
Delft University of Technology, 2016, (de Croon, G.C.H.E. (mentor)).
@mastersthesis{uuid:4b4c4e4b-5e45-4166-bd2c-f35a1e495c6a,
title = {Learning Depth from Single Monocular Images Using Stereo Supervisory Input},
author = {J. Paquim},
url = {http://resolver.tudelft.nl/uuid:4b4c4e4b-5e45-4166-bd2c-f35a1e495c6a},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {Stereo vision systems are often employed in robotics as a means for obstacle avoidance and navigation. These systems have inherent depth-sensing limitations, with significant problems in occluded and untextured regions, leading to sparse depth maps. We propose using a monocular depth estimation algorithm to tackle these problems, in a Self-Supervised Learning (SSL) framework. The algorithm learns online from the sparse depth map generated by a stereo vision system, producing a dense depth map. The algorithm is designed to be computationally efficient, for implementation onboard resource-constrained mobile robots and unmanned aerial vehicles. Within that context, it can be used to provide both reliability against a stereo camera failure, as well as more accurate depth perception, by filling in missing depth information, in occluded and low texture regions. This in turn allows the use of more efficient sparse stereo vision algorithms. We test the algorithm offline on a new, high resolution, stereo dataset, of scenes shot in indoor environments, and processed using both sparse and dense stereo matching algorithms. It is shown that the algorithm’s performance doesn’t deteriorate, and in fact sometimes improves, when learning only from sparse, high confidence regions rather than from the computationally expensive, dense, occlusion-filled and highly post-processed dense depth maps. This makes the approach very promising for self- supervised learning on autonomous robots.},
note = {de Croon, G.C.H.E. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Stereo vision systems are often employed in robotics as a means for obstacle avoidance and navigation. These systems have inherent depth-sensing limitations, with significant problems in occluded and untextured regions, leading to sparse depth maps. We propose using a monocular depth estimation algorithm to tackle these problems, in a Self-Supervised Learning (SSL) framework. The algorithm learns online from the sparse depth map generated by a stereo vision system, producing a dense depth map. The algorithm is designed to be computationally efficient, for implementation onboard resource-constrained mobile robots and unmanned aerial vehicles. Within that context, it can be used to provide both reliability against a stereo camera failure, as well as more accurate depth perception, by filling in missing depth information, in occluded and low texture regions. This in turn allows the use of more efficient sparse stereo vision algorithms. We test the algorithm offline on a new, high resolution, stereo dataset, of scenes shot in indoor environments, and processed using both sparse and dense stereo matching algorithms. It is shown that the algorithm’s performance doesn’t deteriorate, and in fact sometimes improves, when learning only from sparse, high confidence regions rather than from the computationally expensive, dense, occlusion-filled and highly post-processed dense depth maps. This makes the approach very promising for self- supervised learning on autonomous robots.
Goyal, P.
Mission Planning for Sensor Network Deployment using a Fleet of Drones Masters Thesis
Delft University of Technology, 2016, (Hoekstra, J.M. (mentor); Blacquiere, G. (mentor); de Croon, G.C.H.E. (mentor); Smeur, E.J.J. (mentor)).
@mastersthesis{uuid:e5604e9a-c241-4236-83dd-5fc823e7e284,
title = {Mission Planning for Sensor Network Deployment using a Fleet of Drones},
author = {P. Goyal},
url = {http://resolver.tudelft.nl/uuid:e5604e9a-c241-4236-83dd-5fc823e7e284},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {Various methods for route planning of on-road vehicles to serve transportation requests have been developed in the literature in order to reduce transportation and operational costs. The applicability and thus development of these methods is primarily motivated by the field of application. This article deals with the mission planning for a fleet of drones to deploy sensors in a network. In particular, they are conceived to complete the task of delivering geophones in the seismic surveys. Unlike conventional on-road vehicles used for delivery purposes, every drone in the fleet is constrained to make a frequent return trip back to the depot to pick-up a new payload and restore its battery. A centralized planner is proposed in this article due to this constraint. The problem of planning is decomposed into two phases: route formation and route scheduling. The first phase is handled using the extensive formulation of Multi-Trip Vehicle Routing Problem (MTVRP) aiming at minimizing the overall journey time. A heuristic method is also proposed for this phase which provides near-optimal solutions in a computationally efficient manner. The second phase of the planning algorithm deals with the unaddressed problem of depot congestion arising due to the frequent visits of each drone to the depot. This problem is expressed in the form of a Mixed-Integer Linear Program (MILP) that can be solved using available software. This phase is computationally intensive and comparatively slow which restricts the usage of this mission planner in the re-planning phase to the cases involving longer journeys with limited number of routes. The results from a flight-test are also presented in order to demonstrate the mission planner.},
note = {Hoekstra, J.M. (mentor); Blacquiere, G. (mentor); de Croon, G.C.H.E. (mentor); Smeur, E.J.J. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Various methods for route planning of on-road vehicles to serve transportation requests have been developed in the literature in order to reduce transportation and operational costs. The applicability and thus development of these methods is primarily motivated by the field of application. This article deals with the mission planning for a fleet of drones to deploy sensors in a network. In particular, they are conceived to complete the task of delivering geophones in the seismic surveys. Unlike conventional on-road vehicles used for delivery purposes, every drone in the fleet is constrained to make a frequent return trip back to the depot to pick-up a new payload and restore its battery. A centralized planner is proposed in this article due to this constraint. The problem of planning is decomposed into two phases: route formation and route scheduling. The first phase is handled using the extensive formulation of Multi-Trip Vehicle Routing Problem (MTVRP) aiming at minimizing the overall journey time. A heuristic method is also proposed for this phase which provides near-optimal solutions in a computationally efficient manner. The second phase of the planning algorithm deals with the unaddressed problem of depot congestion arising due to the frequent visits of each drone to the depot. This problem is expressed in the form of a Mixed-Integer Linear Program (MILP) that can be solved using available software. This phase is computationally intensive and comparatively slow which restricts the usage of this mission planner in the re-planning phase to the cases involving longer journeys with limited number of routes. The results from a flight-test are also presented in order to demonstrate the mission planner.
Silva, J. P. Rocha
Quadrotor Thrust Vectoring Control with Time Optimal Trajectory Planning in Constant Wind Fields Masters Thesis
Delft University of Technology, 2016, (de Croon, G.C.H.E. (mentor)).
@mastersthesis{uuid:e714d5b9-236d-4509-bc80-79035c0f4725,
title = {Quadrotor Thrust Vectoring Control with Time Optimal Trajectory Planning in Constant Wind Fields},
author = {J. P. Rocha Silva},
url = {http://resolver.tudelft.nl/uuid:e714d5b9-236d-4509-bc80-79035c0f4725},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {This work proposes a control strategy to follow time optimal trajectories planned to visit a given set of waypoints in windy conditions. The aerodynamic effects of quadrotors are investigated, with emphasis on blade flapping, induced and parasitic drag. An extended method to identify all the aerodynamic coefficients is developed, and their influence on the performance is analyzed. A computationally efficient three steps approach is suggested to optimize the trajectory, by minimizing aerodynamic drag and jerk while still guaranteeing near optimal results. The derived smooth trajectory is compared with standard discrete point to point followed by low-pass filtering trajectories, showing energetic improvements in thrust and reductions in Euler angles aggressiveness. By exploiting the non-linear aerodynamic effects and using a priori trajectory information, a thrust vectoring controller is designed and compared with a standard PID controller, showing an increase in performance by reducing the tracking delay and extending the flight envelope.},
note = {de Croon, G.C.H.E. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
This work proposes a control strategy to follow time optimal trajectories planned to visit a given set of waypoints in windy conditions. The aerodynamic effects of quadrotors are investigated, with emphasis on blade flapping, induced and parasitic drag. An extended method to identify all the aerodynamic coefficients is developed, and their influence on the performance is analyzed. A computationally efficient three steps approach is suggested to optimize the trajectory, by minimizing aerodynamic drag and jerk while still guaranteeing near optimal results. The derived smooth trajectory is compared with standard discrete point to point followed by low-pass filtering trajectories, showing energetic improvements in thrust and reductions in Euler angles aggressiveness. By exploiting the non-linear aerodynamic effects and using a priori trajectory information, a thrust vectoring controller is designed and compared with a standard PID controller, showing an increase in performance by reducing the tracking delay and extending the flight envelope.
Coppola, M.
Relative Localization for Collision Avoidance in Micro Air Vehicle Teams Masters Thesis
Delft University of Technology, 2016, (De Croon, G. (mentor)).
@mastersthesis{uuid:8145032a-9b1c-48c1-bf46-6f9d7405e5ef,
title = {Relative Localization for Collision Avoidance in Micro Air Vehicle Teams},
author = {M. Coppola},
url = {http://resolver.tudelft.nl/uuid:8145032a-9b1c-48c1-bf46-6f9d7405e5ef},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {A current limitation of using Micro Air Vehicles in teams is the high risk of collisions between members. Knowledge of relative location is needed in order to perform evasive maneuvers from such collisions. We propose an on-board Bluetooth-based relative localization scheme. Bluetooth is a light-weight and energy efficient communication technology that is readily available on even the smallest Micro Air Vehicle units. In this work, it is exploited for communication between team members to exchange on-board states (velocity, height, and orientation), and the strength of the communication signal is used to infer relative range. The data is fused on-board by each Micro Air Vehicle to obtain a relative estimate of the location and motion of all other team members. Furthermore, a collision avoidance controller is proposed based on collision cones. It is designed to deal with the expected performance of the localization scheme by adapting the collision cones during flight and enforcing a clock-wise evasion maneuver. The system was tested with a team of AR-Drones 2.0 flying in a 4m×4m arena. The task requested the AR-Drones to repeatedly fly from wall to wall whilst passing through the center of the arena, hence making collisions highly likely. The system showed promising results. When using two AR-Drones and off-board velocity/orientation estimates, the drones are able to fly around the arena and avoid each other for the entire flight time as permitted by the battery. With three AR-Drones under the same conditions, flight time to collision was 3 minutes. With two AR-Drones flying with on-board velocity estimation, the time to collision was approximately 3 minutes due to the disturbances in velocity estimates. Simulation results show that significantly better results can be expected with smaller units.},
note = {De Croon, G. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
A current limitation of using Micro Air Vehicles in teams is the high risk of collisions between members. Knowledge of relative location is needed in order to perform evasive maneuvers from such collisions. We propose an on-board Bluetooth-based relative localization scheme. Bluetooth is a light-weight and energy efficient communication technology that is readily available on even the smallest Micro Air Vehicle units. In this work, it is exploited for communication between team members to exchange on-board states (velocity, height, and orientation), and the strength of the communication signal is used to infer relative range. The data is fused on-board by each Micro Air Vehicle to obtain a relative estimate of the location and motion of all other team members. Furthermore, a collision avoidance controller is proposed based on collision cones. It is designed to deal with the expected performance of the localization scheme by adapting the collision cones during flight and enforcing a clock-wise evasion maneuver. The system was tested with a team of AR-Drones 2.0 flying in a 4m×4m arena. The task requested the AR-Drones to repeatedly fly from wall to wall whilst passing through the center of the arena, hence making collisions highly likely. The system showed promising results. When using two AR-Drones and off-board velocity/orientation estimates, the drones are able to fly around the arena and avoid each other for the entire flight time as permitted by the battery. With three AR-Drones under the same conditions, flight time to collision was 3 minutes. With two AR-Drones flying with on-board velocity estimation, the time to collision was approximately 3 minutes due to the disturbances in velocity estimates. Simulation results show that significantly better results can be expected with smaller units.
Slinger, B. J. M. M.
Attitude Control of a Small Helicopter UAV using Incremental Nonlinear Dynamic Inversion Masters Thesis
Delft University of Technology, 2016, (Remes, B.D.W. (mentor)).
@mastersthesis{uuid:9eba1543-6f55-4708-9d68-09446a95d6d4,
title = {Attitude Control of a Small Helicopter UAV using Incremental Nonlinear Dynamic Inversion},
author = {B. J. M. M. Slinger},
url = {http://resolver.tudelft.nl/uuid:9eba1543-6f55-4708-9d68-09446a95d6d4},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {This paper presents an attitude controller for a small helicopter Unmanned Aerial Vehicle (UAV) based on Incremental Nonlinear Dynamic Inversion (INDI). INDI is a sensor-based control method which responds quickly to the commanded input, but also to disturbances. While previous implementations of INDI used a control effectiveness matrix describing effects on rotational accelerations, the implementation presented in this paper uses rotational rates. This is possible with small hingeless-rotor helicopters since the rotational rates are achieved almost immediately, but also the transient is taken into account. By doing so, the matrix contains only constants and the control structure is much simpler. The proposed controller is implemented on a small helicopter which weighs less than 50 grams. The performance of the controller is demonstrated with step responses on roll and heading angles. Also disturbance rejection capabilities are demonstrated. Finally, the controller is deliberately configured incorrectly with wrong control effectiveness and actuator model parameters. A theoretical derivation is provided to predict the effect of incorrect parameters. With experiments, it is demonstrated that the helicopter can be stabilized over a wide range of incorrect values. It is concluded that the demonstrated controller is a suitable choice for small autonomous helicopters.},
note = {Remes, B.D.W. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
This paper presents an attitude controller for a small helicopter Unmanned Aerial Vehicle (UAV) based on Incremental Nonlinear Dynamic Inversion (INDI). INDI is a sensor-based control method which responds quickly to the commanded input, but also to disturbances. While previous implementations of INDI used a control effectiveness matrix describing effects on rotational accelerations, the implementation presented in this paper uses rotational rates. This is possible with small hingeless-rotor helicopters since the rotational rates are achieved almost immediately, but also the transient is taken into account. By doing so, the matrix contains only constants and the control structure is much simpler. The proposed controller is implemented on a small helicopter which weighs less than 50 grams. The performance of the controller is demonstrated with step responses on roll and heading angles. Also disturbance rejection capabilities are demonstrated. Finally, the controller is deliberately configured incorrectly with wrong control effectiveness and actuator model parameters. A theoretical derivation is provided to predict the effect of incorrect parameters. With experiments, it is demonstrated that the helicopter can be stabilized over a wide range of incorrect values. It is concluded that the demonstrated controller is a suitable choice for small autonomous helicopters.
Dalen, G. J. J. Van
Visual Homing for Micro Aerial Vehicles using Scene Familiarity Masters Thesis
Delft University of Technology, 2016, (De Croon, G.C.H.E. (mentor)).
@mastersthesis{uuid:f72c5d11-d959-4778-831d-2abe07945398,
title = {Visual Homing for Micro Aerial Vehicles using Scene Familiarity},
author = {G. J. J. Van Dalen},
url = {http://resolver.tudelft.nl/uuid:f72c5d11-d959-4778-831d-2abe07945398},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {Autonomous navigation is a major challenge in the development of MAVs. When an algorithm has to be efficient, insect intelligence can be a source of inspiration. An elementary navigation task is homing, which means autonomously returning to the initial location. A promising approach makes use of visual familiarity of a route to determine reference headings during homing. In this thesis an existing biological proof of concept based on desert ants is transferred to MAVs. Vision-in-the-loop experiments in different environments are performed, to investigate the viability of scene familiarity for visual navigation. Trained images are used to determine which control actions to take during homing. To determine familiarity, either a database of stored images is kept or an artificial neural network is used. Different image representations are compared in multiple simulated environments. The use of textons for determining familiarity gives the best performance, but HSV color histograms also perform well and are very efficient. It is concluded that to make this method competitive with other visual navigation approaches, route familiarity should be combined with other methods to improve robustness.},
note = {De Croon, G.C.H.E. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Autonomous navigation is a major challenge in the development of MAVs. When an algorithm has to be efficient, insect intelligence can be a source of inspiration. An elementary navigation task is homing, which means autonomously returning to the initial location. A promising approach makes use of visual familiarity of a route to determine reference headings during homing. In this thesis an existing biological proof of concept based on desert ants is transferred to MAVs. Vision-in-the-loop experiments in different environments are performed, to investigate the viability of scene familiarity for visual navigation. Trained images are used to determine which control actions to take during homing. To determine familiarity, either a database of stored images is kept or an artificial neural network is used. Different image representations are compared in multiple simulated environments. The use of textons for determining familiarity gives the best performance, but HSV color histograms also perform well and are very efficient. It is concluded that to make this method competitive with other visual navigation approaches, route familiarity should be combined with other methods to improve robustness.
Nous, C. W. M.
Performance Evaluation in Obstacle Avoidance Masters Thesis
Delft University of Technology, 2016, (De Croon, G.C.H.E. (mentor)).
@mastersthesis{uuid:9ad6db51-5d2b-4680-b250-72b03ccc5fbb,
title = {Performance Evaluation in Obstacle Avoidance},
author = {C. W. M. Nous},
url = {http://resolver.tudelft.nl/uuid:9ad6db51-5d2b-4680-b250-72b03ccc5fbb},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
abstract = {No quantitative procedure currently exists to evaluate the obstacle avoidance capabilities of robotic applications. Such an evaluation method is needed for comparing different methods, but also to determine the operational limits of autonomous systems. This work proposes an evaluation framework which can find such limits. The framework comprises two sets of tests: detection tests and avoidance tests. For each set, environment and performance metrics need to be defined. For detection tests such metrics are well known. For avoidance tests however such metrics are not readily available. Therefore a new set of metrics is proposed. The framework is applied to a UAV that uses stereo vision to detect obstacles and three different algorithms to calculate the avoidance manoeuvre.},
note = {De Croon, G.C.H.E. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
No quantitative procedure currently exists to evaluate the obstacle avoidance capabilities of robotic applications. Such an evaluation method is needed for comparing different methods, but also to determine the operational limits of autonomous systems. This work proposes an evaluation framework which can find such limits. The framework comprises two sets of tests: detection tests and avoidance tests. For each set, environment and performance metrics need to be defined. For detection tests such metrics are well known. For avoidance tests however such metrics are not readily available. Therefore a new set of metrics is proposed. The framework is applied to a UAV that uses stereo vision to detect obstacles and three different algorithms to calculate the avoidance manoeuvre.
Janssen, Y. S.
Reinforcement Learning Policy Approximation by Behavior Trees: using Genetic Algoritms Masters Thesis
Delft University of Technology, 2016, (Scheper, K.Y.W. (mentor)).
@mastersthesis{uuid:f6008da9-d688-4b9f-9880-8d7c3b51a777,
title = {Reinforcement Learning Policy Approximation by Behavior Trees: using Genetic Algoritms},
author = {Y. S. Janssen},
url = {http://resolver.tudelft.nl/uuid:f6008da9-d688-4b9f-9880-8d7c3b51a777},
year = {2016},
date = {2016-01-01},
school = {Delft University of Technology},
note = {Scheper, K.Y.W. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
2015
Lochem, S. Van
Ecological Interface Design for Collaboration of Multiple UAVs in Remote Areas Masters Thesis
Delft University of Technology, 2015, (Borst, C. (mentor); De Croon, G.C.H.E. (mentor); Mulder, M. (mentor); Van Paassen, M.M. (mentor)).
@mastersthesis{uuid:483b3b05-4e72-44a2-840b-a207e06990af,
title = {Ecological Interface Design for Collaboration of Multiple UAVs in Remote Areas},
author = {S. Van Lochem},
url = {http://resolver.tudelft.nl/uuid:483b3b05-4e72-44a2-840b-a207e06990af},
year = {2015},
date = {2015-01-01},
school = {Delft University of Technology},
abstract = {Unmanned Aerial Vehicles (UAVs) can be used to access remote areas that were otherwise inaccessible, for example, for surveillance missions. Collaboration between them can help overcome communication constraints by building airborne relay networks that allow beyond line of sight communication. This research investigates if a single operator can supervise multiple UAVs in a collaborative surveillance task under communication constraints. For this purpose and ecological interface was designed to support operators in their task and to bring flexibility in the system. A human-in-the-loop evaluation study was performed to investigate the successfulness of operators in the control task of such a mission including an analysis of individual components of the interface. It was shown that operators are able to successfully operate surveillance missions under communication- and battery constraints. Participants did however not completely do this without separation conflicts and communication losses, which indicates that the interface lacks elements representing endurance and separation assurance. To an extent the interface design turned out to be scalable, with a few remaining visualizations that still suffered from this problem. More advanced ways of displaying information on request and grouping of select information is thought to offer opportunities to improve ground control interface on this matter.},
note = {Borst, C. (mentor); De Croon, G.C.H.E. (mentor); Mulder, M. (mentor); Van Paassen, M.M. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Unmanned Aerial Vehicles (UAVs) can be used to access remote areas that were otherwise inaccessible, for example, for surveillance missions. Collaboration between them can help overcome communication constraints by building airborne relay networks that allow beyond line of sight communication. This research investigates if a single operator can supervise multiple UAVs in a collaborative surveillance task under communication constraints. For this purpose and ecological interface was designed to support operators in their task and to bring flexibility in the system. A human-in-the-loop evaluation study was performed to investigate the successfulness of operators in the control task of such a mission including an analysis of individual components of the interface. It was shown that operators are able to successfully operate surveillance missions under communication- and battery constraints. Participants did however not completely do this without separation conflicts and communication losses, which indicates that the interface lacks elements representing endurance and separation assurance. To an extent the interface design turned out to be scalable, with a few remaining visualizations that still suffered from this problem. More advanced ways of displaying information on request and grouping of select information is thought to offer opportunities to improve ground control interface on this matter.
Hecke, K. G. Van
Persistent self-supervised learning principle: Study and demonstration on flying robots Masters Thesis
Delft University of Technology, 2015, (De Croon, G.C.H.E. (mentor); Van der Maaten, L.J.P. (mentor); Izzo, D. (mentor); Hennes, D. (mentor)).
@mastersthesis{uuid:b722da02-089f-42a8-a3ea-fb3f5900bcdd,
title = {Persistent self-supervised learning principle: Study and demonstration on flying robots},
author = {K. G. Van Hecke},
url = {http://resolver.tudelft.nl/uuid:b722da02-089f-42a8-a3ea-fb3f5900bcdd},
year = {2015},
date = {2015-01-01},
school = {Delft University of Technology},
abstract = {We introduce, study and demonstrate Persistent Self-Supervised Learning (PSSL), a machine learning method for usage onboard robotic platforms. The PSSL model leverages a standard supervised learning method to simplify the learning problem, but acquires training data in an unsupervised and autonomous manner. Using two platforms, a small multicopter on earth and the space based test bed SPHERES inside the International Space Station , we demonstrate the PSSL principle on a proof of concept problem: learning monocular depth estimation using stereo vision. The robot operates first in a ground truth mode based on the distance perceived by the stereo system, while persistently learning the environment using monocular cues. After the performance of the estimator transcends a ROC quality measure, the robot switches to operation based on the monocular depth estimates. Our results show the viability of the PSSL method, by being able to navigate a room on the basis of learned monocular vision, without collecting any training data beforehand. We identify a major challenge in PSSL caused by a training bias due to behavioral differences in the estimator and the ground truth based operation; however, this is a known problem also for related learning methods such as reinforcement learning. PSSL helps solve this problem by 1) clearly separating the learning problem from the behavior and 2) the possibility to keep learning during estimator behavior.},
note = {De Croon, G.C.H.E. (mentor); Van der Maaten, L.J.P. (mentor); Izzo, D. (mentor); Hennes, D. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
We introduce, study and demonstrate Persistent Self-Supervised Learning (PSSL), a machine learning method for usage onboard robotic platforms. The PSSL model leverages a standard supervised learning method to simplify the learning problem, but acquires training data in an unsupervised and autonomous manner. Using two platforms, a small multicopter on earth and the space based test bed SPHERES inside the International Space Station , we demonstrate the PSSL principle on a proof of concept problem: learning monocular depth estimation using stereo vision. The robot operates first in a ground truth mode based on the distance perceived by the stereo system, while persistently learning the environment using monocular cues. After the performance of the estimator transcends a ROC quality measure, the robot switches to operation based on the monocular depth estimates. Our results show the viability of the PSSL method, by being able to navigate a room on the basis of learned monocular vision, without collecting any training data beforehand. We identify a major challenge in PSSL caused by a training bias due to behavioral differences in the estimator and the ground truth based operation; however, this is a known problem also for related learning methods such as reinforcement learning. PSSL helps solve this problem by 1) clearly separating the learning problem from the behavior and 2) the possibility to keep learning during estimator behavior.
Szabó, T.
Autonomous Collision Avoidance for Swarms of MAVs: Based solely on RSSI measurements Masters Thesis
Delft University of Technology, 2015, (Mulder, J.A. (mentor); de Croon, G.C.H.E. (mentor); de Visser, C.C. (mentor); Scheper, K.Y.W. (mentor); Verhoeven, C.J.M. (mentor)).
@mastersthesis{uuid:3552d27e-6816-4ea3-85f6-4464deb8f1bd,
title = {Autonomous Collision Avoidance for Swarms of MAVs: Based solely on RSSI measurements},
author = {T. Szabó},
url = {http://resolver.tudelft.nl/uuid:3552d27e-6816-4ea3-85f6-4464deb8f1bd},
year = {2015},
date = {2015-01-01},
school = {Delft University of Technology},
abstract = {Swarming is a promising solution for extending the flight time and payload carrying capabilities of Micro Aerial Vehicles (MAVs), where recent years have brought many advancements. These allow MAVs to operate ever more autonomously by tackling problems such as obstacle avoidance and autonomous navigation. A major challenge that still remains, however, is to ensure collision avoidance within the swarm itself. Avoiding collisions with other members of the swarm requires knowledge of their relative positions - typically requiring additional sensors to be carried on-board. Using the signal strength of the MAVs’ communication link provides an alternative method for estimating relative distances between the members of the swarm without requiring need for any additional sensors.},
note = {Mulder, J.A. (mentor); de Croon, G.C.H.E. (mentor); de Visser, C.C. (mentor); Scheper, K.Y.W. (mentor); Verhoeven, C.J.M. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Swarming is a promising solution for extending the flight time and payload carrying capabilities of Micro Aerial Vehicles (MAVs), where recent years have brought many advancements. These allow MAVs to operate ever more autonomously by tackling problems such as obstacle avoidance and autonomous navigation. A major challenge that still remains, however, is to ensure collision avoidance within the swarm itself. Avoiding collisions with other members of the swarm requires knowledge of their relative positions - typically requiring additional sensors to be carried on-board. Using the signal strength of the MAVs’ communication link provides an alternative method for estimating relative distances between the members of the swarm without requiring need for any additional sensors.
Verdugo, M Paz Gomes
Event-based Optical Flow using a Dynamic Vision Sensor for MAV Landing Masters Thesis
Delft University of Technology, Delft, NL, 2015.
@mastersthesis{PazGomesVerdugo2015,
title = {Event-based Optical Flow using a Dynamic Vision Sensor for MAV Landing},
author = {M Paz Gomes Verdugo},
year = {2015},
date = {2015-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
2014
Noyon, Tijs
TU Delft Aerospace Engineering, 2014, (Bijl, Hester (mentor); van Oudheusden, Bas (mentor); Tay, Weebeng (mentor); de Wagter, Christophe (mentor); Delft University of Technology (degree granting institution)).
@mastersthesis{uuid:cfa5fc25-4d67-482f-91d0-c09f6110af81,
title = {The effect of wing deformation on unsteady aerodynamic mechanisms in hovering flapping flight: Numerical study using a three-dimensional immersed boundary method},
author = {Tijs Noyon},
url = {http://resolver.tudelft.nl/uuid:cfa5fc25-4d67-482f-91d0-c09f6110af81},
year = {2014},
date = {2014-01-01},
school = {TU Delft Aerospace Engineering},
abstract = {This study investigated the effect of chord deformation on the unsteady aerodynamic mechanisms found in hovering flapping flight at a Reynolds number of Re = 2002. This was done in order to get a better understanding of the physics involved in flapping flight, which in turn could lead to improved Micro Aerial Vehicle (MAV) designs. A three-dimensional numerical study was performed using an immersed boundary method (IBM) with the discrete forcing approach. The solver was first validated against an experiment by Kim and Gharib (2011).},
note = {Bijl, Hester (mentor); van Oudheusden, Bas (mentor); Tay, Weebeng (mentor); de Wagter, Christophe (mentor); Delft University of Technology (degree granting institution)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
This study investigated the effect of chord deformation on the unsteady aerodynamic mechanisms found in hovering flapping flight at a Reynolds number of Re = 2002. This was done in order to get a better understanding of the physics involved in flapping flight, which in turn could lead to improved Micro Aerial Vehicle (MAV) designs. A three-dimensional numerical study was performed using an immersed boundary method (IBM) with the discrete forcing approach. The solver was first validated against an experiment by Kim and Gharib (2011).
Melis, Johan
Maneuvering Fruit Fly Flight Masters Thesis
TU Delft Aerospace Engineering, 2014, (van Oudheusden, Bas (mentor); Muijres, Florian (mentor); Remes, Bart (mentor); Perçin, Mustafa (mentor); Delft University of Technology (degree granting institution)).
@mastersthesis{uuid:6d92fc28-9fa0-426f-a374-609a9b8c132c,
title = {Maneuvering Fruit Fly Flight},
author = {Johan Melis},
url = {http://resolver.tudelft.nl/uuid:6d92fc28-9fa0-426f-a374-609a9b8c132c},
year = {2014},
date = {2014-01-01},
school = {TU Delft Aerospace Engineering},
abstract = {The goal of the thesis was to establish and validate a model for maneuvering fruit fly flight. Fruit flies are capable of rapidly changing direction and accelerating away from a threat during so-called escape maneuvers. The maneuverability and control of these escape maneuvers are of interest for the development of small unmanned aircraft (Micro Aerial Vehicles) and for the field of neurobiology where the wing kinematic response of fruit flies on visual stimuli is heavily studied.},
note = {van Oudheusden, Bas (mentor); Muijres, Florian (mentor); Remes, Bart (mentor); Perçin, Mustafa (mentor); Delft University of Technology (degree granting institution)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
The goal of the thesis was to establish and validate a model for maneuvering fruit fly flight. Fruit flies are capable of rapidly changing direction and accelerating away from a threat during so-called escape maneuvers. The maneuverability and control of these escape maneuvers are of interest for the development of small unmanned aircraft (Micro Aerial Vehicles) and for the field of neurobiology where the wing kinematic response of fruit flies on visual stimuli is heavily studied.
Scheper, Kirk Y W
Behaviour Trees for Evolutionary Robotics: Reducing the Reality Gap Masters Thesis
Delft University of Technology, Delft, NL, 2014.
@mastersthesis{Scheper2014,
title = {Behaviour Trees for Evolutionary Robotics: Reducing the Reality Gap},
author = {Kirk Y W Scheper},
url = {http://resolver.tudelft.nl/uuid:dde8d42e-590a-465d-abaf-760ec304760f},
year = {2014},
date = {2014-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
2012
Tijmons, S.
Stereo Vision for Flapping Wing MAVs: Design of an Obstacle Avoidance system Masters Thesis
Delft University of Technology, 2012, (Mulder, J.A. (mentor); De Croon, G.C.H.E. (mentor); Van Kampen, E. (mentor); Remes, B.D.W. (mentor)).
@mastersthesis{uuid:f97ac167-38e5-4155-933c-efa2ee179712,
title = {Stereo Vision for Flapping Wing MAVs: Design of an Obstacle Avoidance system},
author = {S. Tijmons},
url = {http://resolver.tudelft.nl/uuid:f97ac167-38e5-4155-933c-efa2ee179712},
year = {2012},
date = {2012-01-01},
school = {Delft University of Technology},
abstract = {In the field of Micro Air Vehicle (MAV) research the use of flapping wings attracts a lot of interest. The potential of flapping wings lies in their efficiency at small scales and their large flight envelope with a single configuration. They have the possibility of performing both energy efficient long distance flights as well as hovering flights. Most studies on Flapping Wing MAVs (FWMAVs) have focused on the design of the airframe and making them able to fly. Currently, the state-of-the-art permits investigation of the necessary autonomous flight capabilities of FWMAVs. Most previous studies have made important preliminary steps by using external cameras or an onboard camera with the FWMAV flying in a modified environment. However, since autonomy is most useful for flight in unknown areas, it will be necessary to use an onboard camera while flying in unmodified environments. Research in this direction has been performed on the DelFly. In particular, the well-known cue of optic flow was found to be rather unreliable for the determination of 3D distances, and it was complemented by a novel visual appearance cue. Since the combination of these cues may still not be sufficient for robust and long-term obstacle avoidance, this study focuses on a different well-known method to extract 3D information on the environment: stereo vision. The potential advantage of stereo vision over optic flow is that it can provide instantaneous distance estimates, implying a reduced dependence on the complex camera movements during flapping flight. The goal is to employ stereo vision in a computationally efficient way in order to achieve obstacle avoidance. The focus of this study is on using heading control for this task. Four main contributions are made: The first contribution comprises an extensive study on literature in the field of computational stereo vision. This research has been done for decades and a lot of methods were developed. These mainly focus on optimizing the quality of the results, while disregarding computational complexity. In this study the focus was on finding one or more time efficient methods that give sufficient quality to perform robust obstacle avoidance. It was concluded that Semi-Global Matching is a good candidate. The second contribution is that for the first time it has been investigated what the requirements are for a stereo vision system to do successful stereo vision-based obstacle avoidance on FWMAVs. In order to achieve accurate stereo vision results, both hardware and software aspects are found to be of importance. FWMAVs can carry only a small amount of payload and therefore there is a large restriction on sensor weight. The third contribution is the development of a systematical way to use the 3D information extracted by the stereo vision algorithm in order to find a guaranteed collision-free flight path. The focus was on dealing with the limited maneuverability of the MAV and the limited view angle of the camera. The fourth contribution is in giving an indication on the usefulness of stereo vision based on multiple experiments. These focus on determining the accuracy of the obstacle detection method as well as on validating the functionality of the obstacle avoidance strategy. The designed system proved to be successful for the task of obstacle avoidance with FWMAVs. The DelFly II successfully avoided the walls in an indoor office space of 7.3×8.2m for more than 72 seconds. This is a considerable improvement over previous monocular solutions. Since even reasonable obstacle detection could be performed for low-textured white walls, the experiments clearly show the potential of stereo vision for obstacle avoidance of FWMAVs. In combination with existing methods for speed and height control the proposed system has the potential of making fully autonomous (flapping wing) MAVs possible.},
note = {Mulder, J.A. (mentor); De Croon, G.C.H.E. (mentor); Van Kampen, E. (mentor); Remes, B.D.W. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
In the field of Micro Air Vehicle (MAV) research the use of flapping wings attracts a lot of interest. The potential of flapping wings lies in their efficiency at small scales and their large flight envelope with a single configuration. They have the possibility of performing both energy efficient long distance flights as well as hovering flights. Most studies on Flapping Wing MAVs (FWMAVs) have focused on the design of the airframe and making them able to fly. Currently, the state-of-the-art permits investigation of the necessary autonomous flight capabilities of FWMAVs. Most previous studies have made important preliminary steps by using external cameras or an onboard camera with the FWMAV flying in a modified environment. However, since autonomy is most useful for flight in unknown areas, it will be necessary to use an onboard camera while flying in unmodified environments. Research in this direction has been performed on the DelFly. In particular, the well-known cue of optic flow was found to be rather unreliable for the determination of 3D distances, and it was complemented by a novel visual appearance cue. Since the combination of these cues may still not be sufficient for robust and long-term obstacle avoidance, this study focuses on a different well-known method to extract 3D information on the environment: stereo vision. The potential advantage of stereo vision over optic flow is that it can provide instantaneous distance estimates, implying a reduced dependence on the complex camera movements during flapping flight. The goal is to employ stereo vision in a computationally efficient way in order to achieve obstacle avoidance. The focus of this study is on using heading control for this task. Four main contributions are made: The first contribution comprises an extensive study on literature in the field of computational stereo vision. This research has been done for decades and a lot of methods were developed. These mainly focus on optimizing the quality of the results, while disregarding computational complexity. In this study the focus was on finding one or more time efficient methods that give sufficient quality to perform robust obstacle avoidance. It was concluded that Semi-Global Matching is a good candidate. The second contribution is that for the first time it has been investigated what the requirements are for a stereo vision system to do successful stereo vision-based obstacle avoidance on FWMAVs. In order to achieve accurate stereo vision results, both hardware and software aspects are found to be of importance. FWMAVs can carry only a small amount of payload and therefore there is a large restriction on sensor weight. The third contribution is the development of a systematical way to use the 3D information extracted by the stereo vision algorithm in order to find a guaranteed collision-free flight path. The focus was on dealing with the limited maneuverability of the MAV and the limited view angle of the camera. The fourth contribution is in giving an indication on the usefulness of stereo vision based on multiple experiments. These focus on determining the accuracy of the obstacle detection method as well as on validating the functionality of the obstacle avoidance strategy. The designed system proved to be successful for the task of obstacle avoidance with FWMAVs. The DelFly II successfully avoided the walls in an indoor office space of 7.3×8.2m for more than 72 seconds. This is a considerable improvement over previous monocular solutions. Since even reasonable obstacle detection could be performed for low-textured white walls, the experiments clearly show the potential of stereo vision for obstacle avoidance of FWMAVs. In combination with existing methods for speed and height control the proposed system has the potential of making fully autonomous (flapping wing) MAVs possible.
Eisma, Jerke
Flow visualization and force measurements on a flapping-wing MAV DelFly II in forward flight configuration Masters Thesis
TU Delft Aerospace Engineering, 2012, (Scarano, Fulvio (mentor); van Oudheusden, Bas (mentor); Perçin, Mustafa (mentor); Remes, Bart (mentor); Delft University of Technology (degree granting institution)).
@mastersthesis{uuid:7d688e57-328c-4d76-990f-e619221feeb4,
title = {Flow visualization and force measurements on a flapping-wing MAV DelFly II in forward flight configuration},
author = {Jerke Eisma},
url = {http://resolver.tudelft.nl/uuid:7d688e57-328c-4d76-990f-e619221feeb4},
year = {2012},
date = {2012-01-01},
school = {TU Delft Aerospace Engineering},
abstract = {Flapping wing flight has attracted increased interest among aerodynamics researchers recently in view of the recent expansion of design efforts in the field of Micro Aerial Vehicles (MAVs). MAVs are given specific attention because of their potential as mobile platforms capable of reconnaissance and gathering intelligence in hazardous and physically inaccessable areas. To achieve these missions, they should be manoevring with ease, staying aloft and propelling themselves efficiently. Conventional means of aerodynamic force generation are found lacking at this point and the apping-wing approach becomes an appealing or even necessary solution. In contrast to the conventional (fixed and rotary wing) force generation mechanisms, apping wing systems take benefit from the unsteady ow effects that are associated to the vortices separating from the wing leading and trailing edges, which create low pressure regions around the wings that lead to the generation of higher lift and thrust.},
note = {Scarano, Fulvio (mentor); van Oudheusden, Bas (mentor); Perçin, Mustafa (mentor); Remes, Bart (mentor); Delft University of Technology (degree granting institution)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Flapping wing flight has attracted increased interest among aerodynamics researchers recently in view of the recent expansion of design efforts in the field of Micro Aerial Vehicles (MAVs). MAVs are given specific attention because of their potential as mobile platforms capable of reconnaissance and gathering intelligence in hazardous and physically inaccessable areas. To achieve these missions, they should be manoevring with ease, staying aloft and propelling themselves efficiently. Conventional means of aerodynamic force generation are found lacking at this point and the apping-wing approach becomes an appealing or even necessary solution. In contrast to the conventional (fixed and rotary wing) force generation mechanisms, apping wing systems take benefit from the unsteady ow effects that are associated to the vortices separating from the wing leading and trailing edges, which create low pressure regions around the wings that lead to the generation of higher lift and thrust.
Mendes, A. S.
Vision-based automatic landing of a quadrotor UAV on a floating platform: A new approach using incremental backstepping Masters Thesis
Delft University of Technology, 2012, (Chu, Q.P. (mentor); Van Kampen, E. (mentor); Mulder, J.A. (mentor); Remes, B.D.W. (mentor)).
@mastersthesis{uuid:f8848255-1831-482a-be21-2da3ca3e50bb,
title = {Vision-based automatic landing of a quadrotor UAV on a floating platform: A new approach using incremental backstepping},
author = {A. S. Mendes},
url = {http://resolver.tudelft.nl/uuid:f8848255-1831-482a-be21-2da3ca3e50bb},
year = {2012},
date = {2012-01-01},
school = {Delft University of Technology},
abstract = {The development of systems that allow unmanned aerial vehicles, known as UAVs, to perform tasks autonomously is a current trend in aerospace research. The specific aim of this thesis is to study and achieve vision-based automatic landing of a quadrotor UAV on a floating platform, a known target that possesses oscillatory behavior. The research contributions to be taken from this study can be divided into two perspectives, as described below. From a theoretical point of view, a design solution is proposed which includes GPS navigation to enable the quadrotor to find the target, and vision-based control to approach and land upon it. From this design, several control-related issues must then be solved, mainly the development of a controller for the autoland mission. To accomplish this control task, an incremental backstepping control law is developed. Additionally, linear and standard backstepping controllers are designed for comparison. The derived control laws require knowledge of the states to close the feedback loops; therefore, state estimation algorithms are designed for complete state reconstruction. The approach selected is modular, thus separating position/velocity estimation from attitude determination. The former is performed using an extended Kalman filter, and the latter using a complementary filter. Furthermore, an augmented Kalman filter formulation is developed for estimation of the platform’s vertical motion. The combination of control and state estimation algorithms is tested in a simulated environment using a simulation tool developed in this study for Monte-Carlo analysis. This tool allows for evaluation of the design not only for the nominal case, but also for random combinations of external conditions. Results show that successful performance is obtained for the nonlinear controllers since the desired criteria is met and the risk of crashing is demonstrated to be residual. Additional tests show that incremental backstepping is, in general, more robust than standard backstepping in the case of model mismatch, even in the presence of state estimation errors. From a practical perspective, the findings are twofold. First, this thesis presents a procedure to experimentally determine the moments of inertia of the quadrotor by using a two-axis motion simulator and a six-component force/torque sensor. The inertia properties are also determined analytically using two modeling approaches: point mass analysis and assumption of simple geometric shapes. The results show that point mass analysis can lead to erroneous inertia estimation deviation of 20-30% from the real value), thus resulting in a significant model mismatch. The experimental and simple shapes assumption methods render similar results, which strongly indicates not only that the experimental method proposed is valid, but also that the assumption of simple geometric shapes can be used as a reliable and cost-effective method to determine moments of inertia of small UAVs. Second, in this thesis the system is tested in real time using an actual quadrotor. Flight tests are performed for hovering above a target with known characteristics, and to achieve this end, a vision system is developed to obtain relative position measurements from images captured by an on-board camera. A Kalman filter is implemented for real-time integration of vision with IMU data, and a linear controller with reference command filters is used. Tuning procedures are then carried out until satisfactory performance is achieved.},
note = {Chu, Q.P. (mentor); Van Kampen, E. (mentor); Mulder, J.A. (mentor); Remes, B.D.W. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
The development of systems that allow unmanned aerial vehicles, known as UAVs, to perform tasks autonomously is a current trend in aerospace research. The specific aim of this thesis is to study and achieve vision-based automatic landing of a quadrotor UAV on a floating platform, a known target that possesses oscillatory behavior. The research contributions to be taken from this study can be divided into two perspectives, as described below. From a theoretical point of view, a design solution is proposed which includes GPS navigation to enable the quadrotor to find the target, and vision-based control to approach and land upon it. From this design, several control-related issues must then be solved, mainly the development of a controller for the autoland mission. To accomplish this control task, an incremental backstepping control law is developed. Additionally, linear and standard backstepping controllers are designed for comparison. The derived control laws require knowledge of the states to close the feedback loops; therefore, state estimation algorithms are designed for complete state reconstruction. The approach selected is modular, thus separating position/velocity estimation from attitude determination. The former is performed using an extended Kalman filter, and the latter using a complementary filter. Furthermore, an augmented Kalman filter formulation is developed for estimation of the platform’s vertical motion. The combination of control and state estimation algorithms is tested in a simulated environment using a simulation tool developed in this study for Monte-Carlo analysis. This tool allows for evaluation of the design not only for the nominal case, but also for random combinations of external conditions. Results show that successful performance is obtained for the nonlinear controllers since the desired criteria is met and the risk of crashing is demonstrated to be residual. Additional tests show that incremental backstepping is, in general, more robust than standard backstepping in the case of model mismatch, even in the presence of state estimation errors. From a practical perspective, the findings are twofold. First, this thesis presents a procedure to experimentally determine the moments of inertia of the quadrotor by using a two-axis motion simulator and a six-component force/torque sensor. The inertia properties are also determined analytically using two modeling approaches: point mass analysis and assumption of simple geometric shapes. The results show that point mass analysis can lead to erroneous inertia estimation deviation of 20-30% from the real value), thus resulting in a significant model mismatch. The experimental and simple shapes assumption methods render similar results, which strongly indicates not only that the experimental method proposed is valid, but also that the assumption of simple geometric shapes can be used as a reliable and cost-effective method to determine moments of inertia of small UAVs. Second, in this thesis the system is tested in real time using an actual quadrotor. Flight tests are performed for hovering above a target with known characteristics, and to achieve this end, a vision system is developed to obtain relative position measurements from images captured by an on-board camera. A Kalman filter is implemented for real-time integration of vision with IMU data, and a linear controller with reference command filters is used. Tuning procedures are then carried out until satisfactory performance is achieved.
Koopmans, Andries J
Delfly Freeflight -- Autonomous Flight of the Delfly in the Wind Tunnel using Low-Cost Sensors Masters Thesis
Delft University of Technology, Delft, NL, 2012.
@mastersthesis{koopmans2012,
title = {Delfly Freeflight -- Autonomous Flight of the Delfly in the Wind Tunnel using Low-Cost Sensors},
author = {Andries J Koopmans},
year = {2012},
date = {2012-01-01},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
2011
Stuiver, M.
Delft University of Technology, 2011, (Bijl, H. (mentor); van Oudheusden, B.W. (mentor); Muijres, F.T. (mentor); Remes, B.D.W. (mentor)).
@mastersthesis{uuid:4b95f124-7ea5-49ea-af65-dfb436d6bdc7,
title = {Bats in Gliding Flight: A comparative wind tunnel investigation of the aerodynamics of gliding bats and a bat inspired gliding wing model},
author = {M. Stuiver},
url = {http://resolver.tudelft.nl/uuid:4b95f124-7ea5-49ea-af65-dfb436d6bdc7},
year = {2011},
date = {2011-01-01},
school = {Delft University of Technology},
abstract = {Due to the high cost of flight, there is a high evolutionary selection pressure for energy efficient flight patterns, such as using external natural forces for soaring or flying intermittently. Some bats at time soar, glide or flap glide. Bounding flight is not possible as their membranous wings will go slack, and soaring is not common amongst bats, as most bats are nocturnal and during night thermals are usually of insufficient strength. From an aerodynamic point of view, gliding flight is less complex than flapping flight, however in bats undulating flight patterns are less observed than in birds. So, why should bats glide? Flight performance studies on live bats have revealed a part of the complexity of hovering and steady flapping flight, but gliding flight in these animals is poorly studied. To get insight in how bats glide and in their gliding flight performance, gliding flight of bats is studied from two points of view; gliding flight of real bats and gliding of a flexible, bat inspired wing model, in a low speed, tiltable wind tunnel. The kinematics of both the bats and the model are filmed by two synchronised high speed cameras, and the flow field in a transverse plane behind the wings is visualized by means of a PIV system. Three medium sized bats Leptonycteris yerbabuenae, are trained to glide at a feeder in the test section of the wind tunnel at a know, fixed glide angle. This known glide angle enables to calculate the aerodynamic forces, which are fixed properties in steady gliding flight. A gliding wing model, based on a bat’s wing, with an adjustable leading edge flap, is designed, build, and tested at different angles of attack. The wing model is tested with both a smooth and a structured top surface to see what the effect of ’turbulators’ can be. Additionally the wing model is mounted onto a balance in order to measure the aerodynamic forces. By means of experiments with the wing model, wake structures of gliding flight can be connected to a single changing morphology parameter to explore the parameter space, and the wake structures can be compared to the wake structures of the gliding bats. The bats are observed to glide for some seconds in the test section, but only the parts of the glides at the feeder where the tip vortex strength and position were stable are analysed. From the PIV data, an average wake is constructed per glide sequence of the bats, and for each leading edge setting and speed combination of the model wing. From the average wake the flight forces and the resulting flight performance properties are derived. The wing model approaches the glide behaviour of the bats. Deploying the leading edge flap increases the span efficiency and the lift coefficient at low angles of attack. Also the structure on top of the wing is beneficial for flight performance at low angles of attack.},
note = {Bijl, H. (mentor); van Oudheusden, B.W. (mentor); Muijres, F.T. (mentor); Remes, B.D.W. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Due to the high cost of flight, there is a high evolutionary selection pressure for energy efficient flight patterns, such as using external natural forces for soaring or flying intermittently. Some bats at time soar, glide or flap glide. Bounding flight is not possible as their membranous wings will go slack, and soaring is not common amongst bats, as most bats are nocturnal and during night thermals are usually of insufficient strength. From an aerodynamic point of view, gliding flight is less complex than flapping flight, however in bats undulating flight patterns are less observed than in birds. So, why should bats glide? Flight performance studies on live bats have revealed a part of the complexity of hovering and steady flapping flight, but gliding flight in these animals is poorly studied. To get insight in how bats glide and in their gliding flight performance, gliding flight of bats is studied from two points of view; gliding flight of real bats and gliding of a flexible, bat inspired wing model, in a low speed, tiltable wind tunnel. The kinematics of both the bats and the model are filmed by two synchronised high speed cameras, and the flow field in a transverse plane behind the wings is visualized by means of a PIV system. Three medium sized bats Leptonycteris yerbabuenae, are trained to glide at a feeder in the test section of the wind tunnel at a know, fixed glide angle. This known glide angle enables to calculate the aerodynamic forces, which are fixed properties in steady gliding flight. A gliding wing model, based on a bat’s wing, with an adjustable leading edge flap, is designed, build, and tested at different angles of attack. The wing model is tested with both a smooth and a structured top surface to see what the effect of ’turbulators’ can be. Additionally the wing model is mounted onto a balance in order to measure the aerodynamic forces. By means of experiments with the wing model, wake structures of gliding flight can be connected to a single changing morphology parameter to explore the parameter space, and the wake structures can be compared to the wake structures of the gliding bats. The bats are observed to glide for some seconds in the test section, but only the parts of the glides at the feeder where the tip vortex strength and position were stable are analysed. From the PIV data, an average wake is constructed per glide sequence of the bats, and for each leading edge setting and speed combination of the model wing. From the average wake the flight forces and the resulting flight performance properties are derived. The wing model approaches the glide behaviour of the bats. Deploying the leading edge flap increases the span efficiency and the lift coefficient at low angles of attack. Also the structure on top of the wing is beneficial for flight performance at low angles of attack.
Hummelink, B. A.
Fixed-Wing UAV Integrated Navigation with Low-Cost IMU/GPS Masters Thesis
Delft University of Technology, 2011, (Chu, Q.P. (mentor); Mulder, J.A. (mentor); De Croon, G.C.H.E. (mentor); De Wagter, C. (mentor)).
@mastersthesis{uuid:291207fc-5ada-435c-be39-8e1436fd96c9,
title = {Fixed-Wing UAV Integrated Navigation with Low-Cost IMU/GPS},
author = {B. A. Hummelink},
url = {http://resolver.tudelft.nl/uuid:291207fc-5ada-435c-be39-8e1436fd96c9},
year = {2011},
date = {2011-01-01},
school = {Delft University of Technology},
abstract = {Today, there is an increase in the use of Unmanned Aerial Vehicles (UAV's), for applications that can be considered dull, dirty or dangerous when compared to those applications of conventional aircraft or helicopters. To further increase the use of UAV's, their navigation filters must be robust and reliable. The trend in current autopilot development is defined by the ever decreasing size of vehicles leading to the creation of miniature Inertial Navigation Systems (INS) with low cost, low grade sensors. Small flying vehicles have fast dynamics requiring higher control rates and higher dynamic ranges with minimal available onboard computational capacities. Sensor and processing limitations have consequences for the achievable navigation performance. This in turn poses limits on the minimal vehicle stability, weather conditions and trajectory smoothness. The most important aspect and thesis goal is to guarantee the navigation filter solution robustness during all flight maneuvers. A navigation filter is an integration algorithm that provides a navigation solution on the vehicle's state vector from sensor data. This thesis focuses on one UAV platform in particular, namely small fixed-wing UAV's. One of the main challenges with designed navigation filters is that they can be theoretically stable but the outcome can sometimes not be used. In practice, the navigation filter outcome can give a diverging solution while theoretically stable. The goal of this thesis is to define the minimal requirements of sensors and other hardware for an INS such that the stabilization requirements posed by the vehicle dynamics and size can be satisfied. With the requirements stated, smaller and more dynamic fixed-wing UAV's can be stabilized based on the integrated navigation solution. The developed observability analysis tool is able to provide a quantitative analysis on the state observability that can be used to analyze different systems or sensor configurations. The observability matrix is composed of the system and observer dynamics. The system dynamics is based on the Inertial Measuring Unit (IMU) prediction of the system states, the observer equations correspond to the observer dynamics. A non-linear local observability analysis has been performed to calculate the observability matrix. The traditional Singular-Values Decomposition (SVD) algorithm provides the singular values of an observability matrix in a decreasing order and indicates the rank of the system. The rank of the observability matrix corresponds to the number of observable system states, the SVD can however not directly link the singular values to the system states. To overcome this problem a different matrix decomposition is used that is able to directly couple the singular values to the system states. This developed matrix decomposition algorithm is based on the QR factorization, called QRsvd. With this algorithm it is possible to quantitatively indicate the observability (degree) of each system state. An analysis into the physical properties of fixed-wing aircraft kinematics resulted in new insight into the movement of flying vehicles. Based on the derived kinematics together with the coupling of an IMU, GPS receiver and fixed-wing aircraft kinematics this resulted in new physical insight. This resulted in three angle correction (AC) equations that can be used as additional attitude/heading angle observers to the conventional IMU/GPS integration. With these three additional observers, the three orientation angles become instantaneously observable. Without the AC equations, a rotational rate constraint is always present to integrate the IMU with GPS. GPS receivers and IMU are separate, self-contained subsystems with different updating frequencies and processing times. Resulting clock differences are called time synchronization errors and result in filter estimation problems. A time synchronization requirement is derived, which is a function of changes in vehicle accelerations and filter innovation. The time synchronization requirement is proportional to the magnitude of the change in vehicle accelerations a and negatively proportional to the magnitude of the identification filter innovation. Vehicles with fast dynamics, like fixed-wing UAV's, can have larger changes in vehicle accelerations magnitude, resulting in a more stringent time synchronization requirement. Based on performed simulations and verification with flight test data, it can be concluded that the improved IMU/GPS filter with AC equations can provide a stable long-term navigation solution with accurate short-term performance, by using (Iterated) Extended Kalman filters. During the performed simulations the position states give the largest source of error, due to the large GPS position uncertainty. For the three orientation angles, the heading angle has a larger identification error compared to the pitch and roll angle. For the orientation angles, the influence of atmospheric wind on the identification performance is minimal except for the heading angle due to the presence of a side-slip angle beta. Coordinate transformations between the Earth, North-East-Down (NED) reference frame F_E and the body-fixed reference frame F_B can be performed using a rotational transformation matrix R_BE. The antisymmetric matrix R_BE holds special properties that can be utilized and fits in the category of Special Orthogonal Lie groups with a dimension of three, called SO(3). Based on SO(3) group properties, a non-linear complementary filter can be constructed that uses this matrix as a single state. The non-linear complementary filter on the SO(3) group, can be used as an alternative to conventional Kalman state identification filters. For (I)EKF the heading angle is the largest source of error of the attitude/heading angles, this is also the case for the SO(3) filter. Differences between the SO(3) filter and (I)EKF are due to two aspects. The SO(3) filter uses constant proportional and integrator gains, where Kalman gain matrices include process and observer uncertainties. The other source of differences can be found in the strong coupling between the individual attitude/heading angles for the non-linear SO(3) filter compared to (I)EKF.},
note = {Chu, Q.P. (mentor); Mulder, J.A. (mentor); De Croon, G.C.H.E. (mentor); De Wagter, C. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Today, there is an increase in the use of Unmanned Aerial Vehicles (UAV's), for applications that can be considered dull, dirty or dangerous when compared to those applications of conventional aircraft or helicopters. To further increase the use of UAV's, their navigation filters must be robust and reliable. The trend in current autopilot development is defined by the ever decreasing size of vehicles leading to the creation of miniature Inertial Navigation Systems (INS) with low cost, low grade sensors. Small flying vehicles have fast dynamics requiring higher control rates and higher dynamic ranges with minimal available onboard computational capacities. Sensor and processing limitations have consequences for the achievable navigation performance. This in turn poses limits on the minimal vehicle stability, weather conditions and trajectory smoothness. The most important aspect and thesis goal is to guarantee the navigation filter solution robustness during all flight maneuvers. A navigation filter is an integration algorithm that provides a navigation solution on the vehicle's state vector from sensor data. This thesis focuses on one UAV platform in particular, namely small fixed-wing UAV's. One of the main challenges with designed navigation filters is that they can be theoretically stable but the outcome can sometimes not be used. In practice, the navigation filter outcome can give a diverging solution while theoretically stable. The goal of this thesis is to define the minimal requirements of sensors and other hardware for an INS such that the stabilization requirements posed by the vehicle dynamics and size can be satisfied. With the requirements stated, smaller and more dynamic fixed-wing UAV's can be stabilized based on the integrated navigation solution. The developed observability analysis tool is able to provide a quantitative analysis on the state observability that can be used to analyze different systems or sensor configurations. The observability matrix is composed of the system and observer dynamics. The system dynamics is based on the Inertial Measuring Unit (IMU) prediction of the system states, the observer equations correspond to the observer dynamics. A non-linear local observability analysis has been performed to calculate the observability matrix. The traditional Singular-Values Decomposition (SVD) algorithm provides the singular values of an observability matrix in a decreasing order and indicates the rank of the system. The rank of the observability matrix corresponds to the number of observable system states, the SVD can however not directly link the singular values to the system states. To overcome this problem a different matrix decomposition is used that is able to directly couple the singular values to the system states. This developed matrix decomposition algorithm is based on the QR factorization, called QRsvd. With this algorithm it is possible to quantitatively indicate the observability (degree) of each system state. An analysis into the physical properties of fixed-wing aircraft kinematics resulted in new insight into the movement of flying vehicles. Based on the derived kinematics together with the coupling of an IMU, GPS receiver and fixed-wing aircraft kinematics this resulted in new physical insight. This resulted in three angle correction (AC) equations that can be used as additional attitude/heading angle observers to the conventional IMU/GPS integration. With these three additional observers, the three orientation angles become instantaneously observable. Without the AC equations, a rotational rate constraint is always present to integrate the IMU with GPS. GPS receivers and IMU are separate, self-contained subsystems with different updating frequencies and processing times. Resulting clock differences are called time synchronization errors and result in filter estimation problems. A time synchronization requirement is derived, which is a function of changes in vehicle accelerations and filter innovation. The time synchronization requirement is proportional to the magnitude of the change in vehicle accelerations a and negatively proportional to the magnitude of the identification filter innovation. Vehicles with fast dynamics, like fixed-wing UAV's, can have larger changes in vehicle accelerations magnitude, resulting in a more stringent time synchronization requirement. Based on performed simulations and verification with flight test data, it can be concluded that the improved IMU/GPS filter with AC equations can provide a stable long-term navigation solution with accurate short-term performance, by using (Iterated) Extended Kalman filters. During the performed simulations the position states give the largest source of error, due to the large GPS position uncertainty. For the three orientation angles, the heading angle has a larger identification error compared to the pitch and roll angle. For the orientation angles, the influence of atmospheric wind on the identification performance is minimal except for the heading angle due to the presence of a side-slip angle beta. Coordinate transformations between the Earth, North-East-Down (NED) reference frame F_E and the body-fixed reference frame F_B can be performed using a rotational transformation matrix R_BE. The antisymmetric matrix R_BE holds special properties that can be utilized and fits in the category of Special Orthogonal Lie groups with a dimension of three, called SO(3). Based on SO(3) group properties, a non-linear complementary filter can be constructed that uses this matrix as a single state. The non-linear complementary filter on the SO(3) group, can be used as an alternative to conventional Kalman state identification filters. For (I)EKF the heading angle is the largest source of error of the attitude/heading angles, this is also the case for the SO(3) filter. Differences between the SO(3) filter and (I)EKF are due to two aspects. The SO(3) filter uses constant proportional and integrator gains, where Kalman gain matrices include process and observer uncertainties. The other source of differences can be found in the strong coupling between the individual attitude/heading angles for the non-linear SO(3) filter compared to (I)EKF.
2010
Groen, M. A.
PIV and force measurements on the flapping-wing MAV DelFly II: An aerodynamic and aeroelastic investigation into vortex development Masters Thesis
Delft University of Technology, 2010, (Bijl, H. (mentor); van Oudheusden, B.W. (mentor); Goosen, J.F.L. (mentor); Remes, B.D.W. (mentor)).
@mastersthesis{uuid:610da696-9202-4f11-ba65-bea67d2edd0b,
title = {PIV and force measurements on the flapping-wing MAV DelFly II: An aerodynamic and aeroelastic investigation into vortex development},
author = {M. A. Groen},
url = {http://resolver.tudelft.nl/uuid:610da696-9202-4f11-ba65-bea67d2edd0b},
year = {2010},
date = {2010-01-01},
school = {Delft University of Technology},
abstract = {Recent years have seen an increasing interest in Micro Air Vehicles (MAVs). MAVs are small (micro sized) aircraft and find their application in a multitude of commercial, industrial and military purposes. To perform their missions MAVs should be small sized, have good manoeuvrability, be well controllable and have a broad flight envelope. When flying in small confinements, the ability to fly at low airspeed and to have good manoeuvrability is critical. One type of MAVs, the flapping-wing MAV, particularly has attractive characteristics for flight in confined spaces. DelFly is a biplane flapping-wing MAV designed and built at Delft University of Technology. DelFly is able to hover and has an onboard camera for observation and vision-based control. For the DelFly project a top-down approach is followed, where from the study of a relative large model experience and theoretical insights can be gained, that can assist to create smaller, functional versions of the DelFly. The ultimate aim of the DelFly project is to improve the design to a very small full autonomous aircraft. For the current experimental investigation, force and flow field measurements were performed on a hovering DelFly II, since this model has a broad flight envelope and proven flight performance. The flow field is studied using particle image velocimetry. Due to the flexible wings there is a strong fluid structure interaction, therefore also the in-flight wing deformation is determined. The aerodynamic mechanism generating forces on the DelFly are related to those found in insect flight. Since leading edge vortices (LEVs) in insect flight are identified as the most important unsteady aerodynamic mechanism enhancing lift generation for insects, the development of these for the DelFly are very interesting. The vortex development is studied for various wings, at various flapping frequencies and at various spanwise positions. For the DelFly wing a conical LEV is developed, starting at out-board spanwise positions, approximately halfway during the translation. This LEV grows larger and is shed along the chord and at this time a new LEV starts to grow at the leading edge. This second LEV is dissipated at the end of the out-stroke during wing rotation, but at the end of the in-stroke this LEV moves above the wings and interacts with the counter-rotating LEV from the mirror wing. Inside the vortex tube a spanwise velocity component out-board is present. The shedding of the initial vortex and start of a second LEV is not completely consistent with LEV development for insect flight (which typically operate at a lower Reynolds number).},
note = {Bijl, H. (mentor); van Oudheusden, B.W. (mentor); Goosen, J.F.L. (mentor); Remes, B.D.W. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Recent years have seen an increasing interest in Micro Air Vehicles (MAVs). MAVs are small (micro sized) aircraft and find their application in a multitude of commercial, industrial and military purposes. To perform their missions MAVs should be small sized, have good manoeuvrability, be well controllable and have a broad flight envelope. When flying in small confinements, the ability to fly at low airspeed and to have good manoeuvrability is critical. One type of MAVs, the flapping-wing MAV, particularly has attractive characteristics for flight in confined spaces. DelFly is a biplane flapping-wing MAV designed and built at Delft University of Technology. DelFly is able to hover and has an onboard camera for observation and vision-based control. For the DelFly project a top-down approach is followed, where from the study of a relative large model experience and theoretical insights can be gained, that can assist to create smaller, functional versions of the DelFly. The ultimate aim of the DelFly project is to improve the design to a very small full autonomous aircraft. For the current experimental investigation, force and flow field measurements were performed on a hovering DelFly II, since this model has a broad flight envelope and proven flight performance. The flow field is studied using particle image velocimetry. Due to the flexible wings there is a strong fluid structure interaction, therefore also the in-flight wing deformation is determined. The aerodynamic mechanism generating forces on the DelFly are related to those found in insect flight. Since leading edge vortices (LEVs) in insect flight are identified as the most important unsteady aerodynamic mechanism enhancing lift generation for insects, the development of these for the DelFly are very interesting. The vortex development is studied for various wings, at various flapping frequencies and at various spanwise positions. For the DelFly wing a conical LEV is developed, starting at out-board spanwise positions, approximately halfway during the translation. This LEV grows larger and is shed along the chord and at this time a new LEV starts to grow at the leading edge. This second LEV is dissipated at the end of the out-stroke during wing rotation, but at the end of the in-stroke this LEV moves above the wings and interacts with the counter-rotating LEV from the mirror wing. Inside the vortex tube a spanwise velocity component out-board is present. The shedding of the initial vortex and start of a second LEV is not completely consistent with LEV development for insect flight (which typically operate at a lower Reynolds number).
Bruggeman, Bart
Improving flight performance of DelFly II in hover by improving wing design and driving mechanism Masters Thesis
Delft University of Technology, 2010.
@mastersthesis{Bruggeman2010,
title = {Improving flight performance of DelFly II in hover by improving wing design and driving mechanism},
author = {Bart Bruggeman},
year = {2010},
date = {2010-01-01},
pages = {123},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
Trips, D
Aerodynamic Design and Optimization of a Long-range Mini-UAV Masters Thesis
Delft University of Technology, Delft, NL, 2010.
@mastersthesis{trips2010,
title = {Aerodynamic Design and Optimization of a Long-range Mini-UAV},
author = {D Trips},
year = {2010},
date = {2010-01-01},
number = {December},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
2009
Verveld, M. J.
Optical Flow Based State Estimation for an Indoor Micro Aerial Vehicle Masters Thesis
Delft University of Technology, 2009, (Mulder, J.A. (mentor); Chu, Q.P. (mentor); de Wagter, C. (mentor)).
@mastersthesis{uuid:fe5ea73c-a85b-4a4b-b5d5-fc01d29b2113,
title = {Optical Flow Based State Estimation for an Indoor Micro Aerial Vehicle},
author = {M. J. Verveld},
url = {http://resolver.tudelft.nl/uuid:fe5ea73c-a85b-4a4b-b5d5-fc01d29b2113},
year = {2009},
date = {2009-01-01},
school = {Delft University of Technology},
abstract = {This work addresses the problem of indoor state estimation for autonomous flying vehicles with an optic flow approach. The paper discusses a sensor configuration using six optic flow sensors of the computer mouse type augmented by a three-axis accelerometer to estimate velocity, rotation, attitude and viewing distances. It is shown that the problem is locally observable for a moving vehicle. A Kalman filter is used to extract these states from the sensor data. The resulting approach is tested in a simulation environment evaluating the performance of three Kalman filter algorithms under various noise conditions. Finally, a prototype of the sensor hardware has been built and tested in a laboratory setup. Paper published: Verveld, M.J., Chu, Q.P., De Wagter, C. and Mulder, J.A. “Optic Flow Based State Estimation for an Indoor Micro Air Vehicle” AIAA Guidance, Navigation, and Control Conference, August 2010, Toronto, Canada AIAA 2010-8209, DOI: 10.2514/6.2010-8209},
note = {Mulder, J.A. (mentor); Chu, Q.P. (mentor); de Wagter, C. (mentor)},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
This work addresses the problem of indoor state estimation for autonomous flying vehicles with an optic flow approach. The paper discusses a sensor configuration using six optic flow sensors of the computer mouse type augmented by a three-axis accelerometer to estimate velocity, rotation, attitude and viewing distances. It is shown that the problem is locally observable for a moving vehicle. A Kalman filter is used to extract these states from the sensor data. The resulting approach is tested in a simulation environment evaluating the performance of three Kalman filter algorithms under various noise conditions. Finally, a prototype of the sensor hardware has been built and tested in a laboratory setup. Paper published: Verveld, M.J., Chu, Q.P., De Wagter, C. and Mulder, J.A. “Optic Flow Based State Estimation for an Indoor Micro Air Vehicle” AIAA Guidance, Navigation, and Control Conference, August 2010, Toronto, Canada AIAA 2010-8209, DOI: 10.2514/6.2010-8209
Clercq, K De
Flow visualization and force measurements on a hovering flapping-wing MAV 'DelFly II' Masters Thesis
Delft University of Technology, Delft, NL, 2009.
@mastersthesis{declerck2009,
title = {Flow visualization and force measurements on a hovering flapping-wing MAV 'DelFly II'},
author = {K De Clercq},
year = {2009},
date = {2009-01-01},
number = {December},
address = {Delft, NL},
school = {Delft University of Technology},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
B.Sc. Thesis Assignments at TU Delft
Every year the MAVLab also guides one or more 3rd-year DSE Projects (Design and Synthesis Exercises). Students who join this course have the opportunity of being assigned to such projects.
https://mavlab.tudelft.nl/hybrid-mav-wins-1st-prize-dse-symposium/