Courses

@mastersthesis{uuid:bd671c3b-afc3-4215-a724-dd69512f4715,

title = {Adaptive dynamic incremental nonlinear control allocation: An actuator fault-tolerant control solution for high-performance aircraft},

author = {Noah Stam},

url = {http://resolver.tudelft.nl/uuid:bd671c3b-afc3-4215-a724-dd69512f4715},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Neglecting actuator dynamics in nonlinear control and control allocation can lead to performance degradation, especially when considering fast dynamic systems. This thesis provides a novel method to account for actuator dynamics in the control allocation solution, dynamic incremental nonlinear control allocation, or D-INCA. The incremental approach allows for the implementation of a first order discrete-time actuator dynamics model in the quadratic programming (QP) solver. This model is used to find the optimal command inputs in addition to the desired physical actuator deflections, hereby compensating for actuator dynamics delays. Whereas, the baseline incremental nonlinear control allocation (INCA) approach requires pseudo-control hedging of the outer loop reference to increase closed loop stability margins under actuator dynamics delays. To its advantage, D-INCA does not require feedback of higher order output derivatives than INCA and can be used with nonlinear non-control affine systems. Furthermore, with adaptive D-INCA, or AD-INCA, an actuator dynamics parameter estimator is introduced to adapt the actuator model online, minimizing actuator tracking errors after actuator failures. The proposed methods are applied to a fighter aircraft model with an over-actuated innovative control effectors suite and results are compared to the baseline INCA controller.},

note = {de Visser, C.C. (mentor); Smeur, E.J.J. (graduation committee); Mooij, E. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:b4f643b2-a64f-4fb4-a18f-5012364f7b0f,

title = {Optical Flow Determination using Neuromorphic Hardware with Integrate & Fire Neurons},

author = {Francesco Branca},

url = {http://resolver.tudelft.nl/uuid:b4f643b2-a64f-4fb4-a18f-5012364f7b0f},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Spiking neural networks implemented for sensing and control of robots have the potential to achieve lower latency and power consumption by processing information sparsely and asynchronously. They have been used on neuromorphic devices to estimate optical flow for micro air vehicles navigation, however robotic implementations have been limited to hardware setups with sensing and processing as separate systems. This article investigates a new approach for training a spiking neural network for optical flow to be deployed on the speck2e device from Synsense. The method takes into account the restrictions of the speck2e in terms of network architecture, neuron model, and number of synaptic operations and it involves training a recurrent neural network with ReLU activation functions, which is subsequently converted into a spiking network. A system of weight rescaling is applied after conversion, to ensure optimal information flow between the layers. Our study shows that it is possible to estimate optical flow with Integrate-and-Fire neurons. However, currently, the optical flow estimation performance is still hampered by the number of synaptic operations. As a result, the network presented in this work is able to estimate optical flow in a range of [-4, 1] pixel/s.},

note = {de Croon, G.C.H.E. (mentor); Hagenaars, J.J. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:6a74fb80-425c-4366-8110-fecfb4a1a5fc,

title = {Novel Neuromorphic Hardware Inspired by the Olfactory Pathway Model of the \textit{Drosophila}: Leveraging bio-plausible computational primitives in digital circuits for spatio-temporal processing},

author = {Lyana Usa},

url = {http://resolver.tudelft.nl/uuid:6a74fb80-425c-4366-8110-fecfb4a1a5fc},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Electrical Engineering, Mathematics and Computer Science},

abstract = {Olfactory learning in \textit{Drosophila }larvae exemplifies efficient neural processing in a small-scale network with minimal power consumption. This system enables larvae to anticipate important outcomes based on new and familiar odor stimuli, a process crucial for survival and adaptation. Central to this learning mechanism is the olfactory pathway model, which embodies the principles of synaptic plasticity and associative learning through prediction error coding mediated by specific neuromodulating neurons in the mushroom body, like dopaminergic neurons. There is a pressing need to develop novel computational frameworks that capture the spatio-temporal processes while remaining compatible with the constraints of small-scale neural networks. These frameworks should draw inspiration from the biophysical properties of neurons within the olfactory pathway model, enabling accurate emulation of neural dynamics and efficient learning processes using spiking neural networks. This thesis proposes a framework based on a phenomenological conductance-based leaky integrate-and-fire (COBALIF) neuron model, inspired by the olfactory pathway model of \textit{Drosophila} larvae. By first prototyping the spiking neural network in Intel's Lava Python-based framework, we validated the design on a neuron and system level for a neuromorphic hardware implementation. This was the foundation of a programmable, neuromorphic FPGA architecture capable of adaptive optimization, employed on a Zynq 7000 SoC FPGA. By implementing this architecture in a single-precision floating-point format, we model the real-time neural dynamics of the COBALIF neuron in one-tenth of a millisecond precision. Moreover, our FPGA implementation serves as a feasible prototype for deploying such biologically inspired neurons and their spatio-temporal dependencies in digital design, paving the way for scaling up to small-scale networks.},

note = {Frenkel, C. (mentor); Makinwa, K.A.A. (graduation committee); de Croon, G.C.H.E. (graduation committee); Nawrot, M. P. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:491a7744-dc49-405c-ba38-5198b3e839a8,

title = {Towards a Robust Wireless Real-Time Ecological Monitoring System},

author = {Luke Waal},

url = {http://resolver.tudelft.nl/uuid:491a7744-dc49-405c-ba38-5198b3e839a8},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Aerospace Engineering},

abstract = {Climate change poses a serious threat to ecosystems and increases the need for accurate and rigorous monitoring of ecosystems. Current monitoring solutions are often bulky, expensive, and lack critical functionalities such as on-board inference capabilities, robust wireless connections, and a diverse sensor suite. Ecological monitoring projects often suffer from inefficiencies caused by the large time delays between collecting data and analyzing said data, as well as having to spend large amounts of time in the field setting up the sensors manually. This thesis addresses many of these issues by designing a sensor with an extensive sensor suite, robust wireless capabilities and an on-board audio classifier able to perform real-time inference. Furthermore, attention is paid to making the system extendable in the future and allow for potentially integrating the sensors with a drone delivery- and retrieval system. The system tests performed indicate that the system has great potential given more time to tweak some of its identified shortcomings.},

note = {Hamaza, S. (mentor); Rajan, R.T. (mentor); Smeur, E.J.J. (graduation committee); Hendriks, R.C. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:3193131c-6b68-46a2-afe1-964a044dd6f9,

title = {Implications of Propeller-Wing Interactions on the Control of Aerodynamic-Surface-Free Tilt-Rotor Quad-Planes},

author = {Noah Wechtler},

url = {http://resolver.tudelft.nl/uuid:3193131c-6b68-46a2-afe1-964a044dd6f9},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Quad-planes are a type of vehicle which combine the hovering capability of quadcopters and the forward flight efficiency of winged aircraft. Flight tests conducted on a dual-axis tilting-rotor quad-plane, designed to fly without aerodynamic surfaces, observed that the quad-plane suffered from insufficient roll authority during fast, forward flight. Subsequent wind tunnel testing confirmed a two- to fourfold reduction in roll moment generation from propellers mounted in front of the wing at similar levels of tilt as their rear counterparts, caused by propeller-wing interactions. To address the mismatch in actuator effectiveness shown by the wind tunnel experiment, the effect of the propeller-wing interactions was incorporated into the aero-propulsive model by means of a global polynomial, the structure of which was found using multivariate orthogonal function modelling. New flight tests demonstrated that, by including the propeller-wing interactions in the control allocation, the vehicle is capable of tracking a figure 8 maneuver without aerodynamic surfaces, and without compromising tracking performance.},

note = {Smeur, E.J.J. (mentor); Mancinelli, A. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:eb522c6b-1b1d-4988-8a7a-e2846dc697c5,

title = {Self-Supervised Learning of Event-Based Optical Flow via Deep Equilibrium Models},

author = {Aleksandar Shokolarov},

url = {http://resolver.tudelft.nl/uuid:eb522c6b-1b1d-4988-8a7a-e2846dc697c5},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {The estimation of optical flow, which determines the movement of objects in a visual scene, is a crucial problem in computer vision. It is essential for applications such as autonomous navigation, where precise motion estimation is critical for performance and safety.<br/><br/>Frame-based cameras capture sequences of still images at regular intervals, from which optical flow is traditionally extracted using optimization-based or learning-based methods. Recently, event-based cameras, which detect changes in pixel brightness asynchronously, have gained traction due to their high temporal resolution and robustness to motion blur, and many algorithms have been developed to estimate optical flow from this data. IDNet is a learning-based approach that achieves state-of-the-art performance. However, IDNet and similar models face two major challenges: they require labeled ground-truth data for training, which is scarce and difficult to collect, and they rely on recurrent neural networks (RNNs) with a fixed number of refinement iterations. This fixed iteration scheme does not adapt to scene complexity, limiting accuracy for complex flows and increasing computational effort for simpler patterns.<br/><br/>The aim of this project is to explore, implement, and evaluate potential methods to address these two mentioned limitations and enhance the capabilities of models like IDNet.<br/><br/>To remove the need for ground-truth data, a self-supervised learning paradigm was implemented by introducing a novel contrast maximization loss that assesses the blur present when accumulating raw events for a certain time interval and compensating it with the predicted flow. To assess the effectiveness of this method, models were trained on the benchmark MVSEC dataset, showing improved results over previous methods with up to 15% on some sequences and an 8% improvement on average. Based on these experiments and results, further research directions were proposed.<br/><br/>As for the problem of the current fixed iteration scheme, Deep Equilibrium Models were found to provide a promising pathway to solving it. These novel models reformulate their iterative structure into a root-finding problem and utilize traditional solvers to find a solution based on some tolerance, providing a trade-off between speed and accuracy. Moreover, they allow for direct differentiation through the network using only their final estimate, compared to previous methods that keep track of their state through all iterations, leading to an O(1) memory consumption. Implementing these and some additional ideas, the trained DEQ IDNet model reached competitive performance on the DSEC dataset while consuming 15% less memory. Yet, further work is needed to close the gap and achieve state-of-the-art performance.},

note = {de Croon, G.C.H.E. (graduation committee); Wu, Y. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:7e9dd0aa-4d24-4100-a224-14e71f86cdda,

title = {IUVO: An Emergency Response Flyer},

author = {Gabriel Gervas Montoya},

url = {http://resolver.tudelft.nl/uuid:7e9dd0aa-4d24-4100-a224-14e71f86cdda},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering; Grochowski, Bartłomiej},

note = {Smeur, E.J.J. (mentor); Varriale, Carmine (graduation committee); Georgopoulos, P. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:8649d62f-6266-44a4-89de-5b5805d83ae5,

title = {Deep Learning Fusion of Monocular and Stereo Depth Maps Using Convolutional Neural Networks},

author = {Dani Tóth},

url = {http://resolver.tudelft.nl/uuid:8649d62f-6266-44a4-89de-5b5805d83ae5},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {This paper presents an encoder-decoder-style convolutional neural network (CNN) for the purpose of improving monocular and stereo depth estimation (SDE) estimates, by combining them with the corresponding monocular estimates through a fusion network, assisted by prior information to provide context for the fusion. Video cameras are commonly used for depth perception in robotics, especially weight-sensitive applications, such as on Micro Aerial Vehicles (MAV). The two primary paradigms for vision-based depth perception are monocular and stereo depth or disparity estimation, each having their own strengths and weaknesses. These strengths and weaknesses seem to be complementary, and thus a fusion of the two may result in more accurate predictions. In this paper, we investigate this fusion by training a CNN that combines stereo and monocular depth or disparity estimates. The fusion network is agnostic to the choice of the input networks, providing great flexibility. It was found that such a fusion network, while increasing the computational complexity of the depth perception pipeline, indeed improves the accuracy of the estimates. The number of outlier predictions has been significantly decreased, while also limiting some fundamental limitations of both stereo and monocular methods, such as errors arising from occluded regions.},

note = {de Croon, G.C.H.E. (mentor); van Dijk, Tom (mentor); de Wagter, C. (graduation committee); Eleftheroglou, N. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:5aeb0475-b3d5-45a4-82c8-7b92fabbb683,

title = {Fault Tolerant Control in Over-Actuated Hybrid Tilt-Rotor Unmanned Aerial Vehicles},

author = {Nico Voß},

url = {http://resolver.tudelft.nl/uuid:5aeb0475-b3d5-45a4-82c8-7b92fabbb683},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Quad-planes combine hovering and vertical takeoff and landing capability with fast and efficient forward flight. Regular Quad-planes with dedicated pusher motor can be subject to gust disturbances, and are not well-equipped to deal with actuator faults. Dual-axis Tilt-Rotor quad-planes are more maneuverable due to their overactuation. This also increases their gust resilience and allows them to hover statically after actuator failures. The vehicle in this paper uses an Incremental Nonlinear Dynamic Inversion (INDI ) controller, combined with a nonlinear Sequential Quadratic Programming (SQP) Control Allocation (CA ) algorithm, which can also find hover solutions in the case of actuator failures. We investigate both a combined allocation of linear and angular accelerations, as well as a cascaded allocation scheme. Due to large required changes in roll and pitch angles, the cascaded approach is selected in this research. Introduction of a tertiary control effort term, separation of attitude and actuator command optimization and a simulated Fault Detection and Identification ( FDI) mechanism led to repeated successful recovery from a motor failure in hover. Position tracking was demonstrated under failure in the recon- figured flight condition. Index Terms- Tilt-rotor, dual-axis tilt, quad-plane, FTC, over- actuated, control allocation},

note = {Smeur, E.J.J. (mentor); Mancinelli, A. (mentor); Bombelli, A. (graduation committee); de Visser, C.C. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:9ab2b4ba-8f91-4891-8190-4a96f77c471e,

title = {Adaptive Visual Servoing Control for Quadrotors: A Bio-inspired Strategy Using Active Vision},

author = {Sander Hazelaar},

url = {http://resolver.tudelft.nl/uuid:9ab2b4ba-8f91-4891-8190-4a96f77c471e},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {New insights into the landing behavior of bumblebees show an adaptive strategy where the optical flow expansion of the landing target is step-wise regulated. In this article, the potential benefits of this approach are studied by replicating the landing experiment with a quadrotor. To this end, an open-loop switching method is developed, enabling fast steps in divergence. An adaptive control law is used to deal with non-linear system dynamics, where the control gain is scheduled based on the control effectiveness of the actuator inputs during the steps. It is demonstrated that the quadrotor can reliably land on the target from varying initial positions, and the switching strategy shows a slight reduction in landing time compared to a constant divergence strategy with the same average divergence over distance. This strategy also reduces the maximum velocity during the landing.},

note = {de Croon, G.C.H.E. (mentor); Yedutenko, M. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:9a295d44-1e95-4911-a4a2-4a96c498fe79,

title = {Design of an inherently fully dynamically balanced aerial manipulator with omnidirectional workspace},

author = {Alexander Bom},

url = {http://resolver.tudelft.nl/uuid:9a295d44-1e95-4911-a4a2-4a96c498fe79},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Mechanical, Maritime and Materials Engineering},

abstract = {Drones are increasingly used nowadays, primarily for visual inspection tasks facilitated by onboard cameras. The field of aerial manipulation tries to expand the capabilities of drones by attaching a manipulator, enabling physical interaction. Unfortunately, the usability of aerial manipulators is hindered by disturbances resulting from the movements of the manipulator. These disturbances, including reaction forces and a shifting centre of mass, not only affect manipulation accuracy but also pose safety risks by potentially destabilizing the drone. In this thesis, a design is presented that addresses this challenge by leveraging the theory of dynamic balance. <br/>A new design approach of making a manipulator fly, instead of the common approach of mounting a manipulator arm to a drone was used. This new approach avoids interference with the drone's components, allowing to focus on the design of the manipulator arm. Furthermore, it made it possible to create a manipulator which can manipulate above, to the side and underneath itself. This makes the presented manipulator arm more versatile than common aerial manipulators whose workspace is mostly located only above or below the drone. The kinematics, workspace and balance conditions of the manipulator arm are presented. Furthermore, the design's workspace is optimised while the mass of the manipulator is minimized in a bilevel optimisation. Finally, the design is validated both by simulation and measurements performed with the built prototype.<br/>The design presented is the first inherently fully dynamically balanced manipulator with omnidirectional workspace which can be used for aerial manipulation.<br},

note = {van der Wijk, V. (mentor); Hamaza, S. (mentor); Herder, J.L. (graduation committee); Goosen, J.F.L. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:a164abc6-0103-4fa0-b7ac-c15bce2bce64,

title = {A lightweight quadrotor autonomy system: To navigate in densely cluttered forest environments},

author = {Andreas Zwanenburg},

url = {http://resolver.tudelft.nl/uuid:a164abc6-0103-4fa0-b7ac-c15bce2bce64},

year  = {2024},

date = {2024-01-01},

school = {TU Delft Mechanical, Maritime and Materials Engineering},

abstract = {These days, people see more and more applications for drones, including monitoring rainforests to protect plant and animal species. However, drones face challenges when navigating through the dense and cluttered vegetation of the forest. These environments necessitate advanced autonomous detection and navigation to make the drone traverse robustly and fly safely. In addition, the forest brings extra challenges, such as blocked signals for GPS localisation, remote control, and remote supervising.<br/><br/>In this thesis project, a drone is designed, built, and programmed to navigate autonomously in the rainforest with complete onboard computing and no GPS localisation. This 500-gram drone is being extensively tested and optimized in real forest conditions, and a dataset is being created from its autonomous flights to simulate various configurations of the path-planning algorithm. The results of these simulations on this dataset are then used for thorough research on how the algorithm can downscale to smaller systems and how this affects performance.<br/><br/>By using the results of this research on downscaling, a 100-gram drone is built and programmed to fly in forest conditions with complete onboard computation. Challenging on this small-size drone is the use of low-quality lightweight sensors and processor. The processor only weighs 10 grams, and the depth camera weighs 8 grams. Unique on this small drone is the 3D path planning fully computed onboard and the implementation of a new type of depth camera.},

note = {Wisse, M. (mentor); Hamaza, S. (graduation committee); Benders, D. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:90916a92-95bc-44eb-889e-81555ddd494f,

title = {Microphones as Airspeed Sensors for Micro Air Vehicles},

author = {Momchil Makaveev},

url = {http://resolver.tudelft.nl/uuid:90916a92-95bc-44eb-889e-81555ddd494f},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {This project proposes and evaluates a novel concept for an airspeed instrument aimed at small hybrid unmanned aerial vehicles. The working principle is to relate the power spectra of the wall-pressure fluctuations beneath the turbulent boundary layer formed over the vehicle’s body to its airspeed. The instrument consists of two microphones, flush mounted on the UAV’s nose cone, that capture the pseudo-sound caused by the coherent turbulent structures, and a micro-controller that processes the signals from the microphones and computes the airspeed. Dedicated models were constructed, using data obtained from wind tunnel and flight experiments, that take the power spectra of the microphones’ signals as an input and provide the airspeed as an output. The model structure is a feed-forward neural network with a single hidden layer, trained using a second-order gradient descent algorithm, following a supervised learning approach. The models were validated using only flight data, with the best one achieving a mean approximation error of 0.043 m/s and having a standard deviation of 1.039 m/s. It was also shown that the airspeed could be successfully predicted for a wide range of angles of attack, given that they are known, thus necessitating the vehicle to be equipped with a dedicated angle of attack sensor.},

note = {Smeur, E.J.J. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:f5c8a6b4-a43b-43f1-9a2d-72d850515369,

title = {Frustumbug: a 3D Mapless Stereo-Vision-based Bug Algorithm for Micro Air Vehicles},

author = {Ruben Meester},

url = {http://resolver.tudelft.nl/uuid:f5c8a6b4-a43b-43f1-9a2d-72d850515369},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {We present a computationally cheap 3D bug algorithm for drones, using stereo vision. Obstacle avoidance is important, but difficult for robots with limited resources, such as drones. Stereo vision requires less weight and power than active distance measurement sensors, but typically has a limited Field of View (FoV). In addition, the stereo camera is fixed on the drone, preventing sensor movement. For obstacle avoidance, bug algorithms require few resources. We base our proposed algorithm, Frustumbug, on the Wedgebug algorithm, since this bug algorithm copes with a limited FoV. Since Wedgebug only focuses on 2D problems, the Local-epsilon-Tangent-Graph (LETG) is used to extend the path planning to 3D. Disparity images are obtained through an optimised stereo block matching algorithm. Obstacles are expanded in disparity space to obtain the configuration space. Furthermore, Frustumbug has an improved robustness to noisy range sensor data, and includes reversing, climbing and descending manoeuvres to avoid or escape local minima. The algorithm has been extensively tested with 225 flights in two challenging simulated environments, with a success rate of 96%. Here, 3.6% did not reach the goal and 0.4% collided. Frustumbug has been implemented on a 20 gram stereo vision system, and guides drones safely around obstacles in the real world, showing its potential for small drones to reach their targets fully autonomously.},

note = {de Croon, G.C.H.E. (mentor); van Dijk, Tom (mentor); de Wagter, C. (graduation committee); Verhoeven, C.J.M. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:3e922cc4-83a0-43aa-a223-a67e554d2e92,

title = {Deep Vision-based Relative Localisation by Monocular Drones},

author = {Raoul Mink},

url = {http://resolver.tudelft.nl/uuid:3e922cc4-83a0-43aa-a223-a67e554d2e92},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Decentralised drone swarms need real time collision avoidance, thus requiring efficient, real time relative localisation. This paper explores different data inputs for vision based relative localisation. It introduces a novel dataset generated in \textit{Blender}, providing ground truth optic flow and depth. Comparisons to \textit{MPI Sintel}, an industry/research standard optic flow dataset, show it to be a challenging and realistic dataset. Two Deep Neural Network (DNN) architectures (YOLOv3 & U-Net) were trained on this data, comparing optic flow to colour images for relative positioning. The results indicate that using optic flow provides a significant advantage in relative localisation. The flow based YOLOv3 had an mAP of 48%, 9% better than the RGB based YOLOv3, and 23% better than its equivalent U-Net. Its IoU_{0.5} of 63% was also 14% better than the RGB based YOLOv3, and 51% than its equivalent U-Net. As an input, it generalises better than RGB, as test clips with variant drones show. For these variants, the optical flow based networks outperformed the RGB based networks by a factor of 10.},

note = {de Croon, G.C.H.E. (mentor); de Wagter, C. (graduation committee); Zarouchas, D. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:bffb47bf-5864-4b18-921b-588b3a664866,

title = {Computationally Efficient Control Allocation Using Active-Set Algorithms},

author = {Till Blaha},

url = {http://resolver.tudelft.nl/uuid:bffb47bf-5864-4b18-921b-588b3a664866},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {An effective distribution of flight control commands over many aircraft actuators (engines, control surfaces, flaps, etc.) can be achieved with constrained optimisation. Active-Set methods solve these problems efficiently, but their computational time requirements are still prohibitive for aircraft with many actuators or slower digital flight control processors. This work shows how these methods can be improved in these regards, by updating the required matrix factorisations at lower computational costs, rather than solving a separate optimisation problem at every step of the iterative algorithm. Additionally, it is shown how the sparsity of the problem matrices can be exploited. Both open-loop simulations and flight tests have been performed, which show that worst-case timings for a 6-rotor multicopter UAV can be improved by 65% over a current Active-Set solver. Furthermore, methods are presented that remedy numerical stability issues occurring in micro-controller floating point arithmetic but introduce a small but measurable adverse effect on the flight behaviour.},

note = {Smeur, E.J.J. (mentor); Remes, B.D.W. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:a4f3199c-71f6-4182-bd98-30db62db8018,

title = {Stability control and positional water jet placement for a novel tethered unmanned hydro-propelled aerial vehicle using real-time water jet detection},

author = {Xander Beurden},

url = {http://resolver.tudelft.nl/uuid:a4f3199c-71f6-4182-bd98-30db62db8018},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Aerial platforms designed for water jet placement are gaining interest in the areas of fire-fighting, washing, and irrigation. A novel, lightweight, and simplistic design is proposed that reduces the number of actuators and limits ineffective water discharge. External camera feedback was used for position control as a first step towards autonomous flight. An initial prototype of an unmanned hydro-propelled aerial vehicle (UHAV) connected to a water hose was designed and fabricated. Flight tests were conducted to show that attitude control with uniaxial thrust-vectoring of two nozzles was impossible due to undamped vibrations and coupling effects. By redesigning the PID controller, pitch rate damping was accomplished. Furthermore, a design trade-off led to the introduction of a canting keel to reduce bank-yaw coupling effects due to asymmetric nozzle deflections. Flight tests proved that the iterated design with a hose length of 3m was capable of disturbance rejection and setpoint tracking. An external camera was used to show that the Lucas-Kanade optical flow algorithm and the implementation of the YOLOv5 segmentation model can be used for positional water jet placement. By increasing the pitch rate damping, improving the water jet detection algorithm and implementing a cost function for water discharge at the area of interest, autonomous missions can be flown in the future.},

note = {de Wagter, C. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:35f8de9d-98a3-457e-8f02-33d2a40de595,

title = {Design of an Aerial-Aquatic Inspection Drone},

author = {Louis Ronsse},

url = {http://resolver.tudelft.nl/uuid:35f8de9d-98a3-457e-8f02-33d2a40de595},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering; De Vusser, Mathis},

note = {Hamaza, S. (mentor); Dransfeld, C.A. (mentor); van Oosterom, S.J.M. (graduation committee); Jigjid, K. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:99e01573-9ec5-4c67-8b65-e266ad83098a,

title = {Maritime Drone Swarm},

author = {Doruk Çelebi},

url = {http://resolver.tudelft.nl/uuid:99e01573-9ec5-4c67-8b65-e266ad83098a},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering; de Bruijn, Marnix},

note = {Remes, B.D.W. (mentor); Giovanardi, Bianca (graduation committee); Westerbeek, S.H.J. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:1edec476-3b58-458d-a4a6-cbba30b783e6,

title = {Evolving Spiking Neural Networks to Mimic PID Control: Applied to Autonomous Blimps},

author = {Tim Burgers},

url = {http://resolver.tudelft.nl/uuid:1edec476-3b58-458d-a4a6-cbba30b783e6},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {In recent years, Artificial Neural Networks (ANN) have become a standard in robotic control. However, a significant drawback of large-scale ANNs is their increased power consumption. This becomes a critical concern when designing autonomous aerial vehicles, given the stringent constraints on power and weight. Especially in the case of blimps, known for their extended endurance, power-efficient control methods are essential. Spiking neural networks (SNN) can provide a solution, facilitating energy-efficient and asynchronous event-driven processing. <br/>In this paper, we have evolved SNNs for accurate altitude control of a non-neutrally buoyant indoor blimp, relying solely on onboard sensing and processing power. The blimp's altitude tracking performance significantly improved compared to prior research, showing reduced oscillations and a minimal steady-state error. The parameters of the SNNs were optimized via an evolutionary algorithm, using a Proportional-Derivative-Integral (PID) controller as the target signal. We developed two complementary SNN controllers while examining various hidden layer structures. The first controller responds swiftly to control errors, mitigating overshooting and oscillations, while the second minimizes steady-state errors due to non-neutral buoyancy-induced drift. Despite the blimp's drivetrain limitations, our SNN controllers ensured stable altitude control, employing only 160 spiking neurons.},

note = {de Croon, G.C.H.E. (mentor); Stroobants, S. (mentor); de Wagter, C. (graduation committee); Bombelli, A. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:21b4203a-abbb-47d7-bbd5-df042d8d7b53,

title = {Vision-guided Quadrotor Perching on Imperfectly Cylindrical Structures},

author = {Seamus McGinley},

url = {http://resolver.tudelft.nl/uuid:21b4203a-abbb-47d7-bbd5-df042d8d7b53},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {The design of aerial robots capable of perching poses significant challenges, from requiring pilots to master precise manoeuvres, to devising hardware and software capable of adapting to diverse perch structures and complex field environments. The Slapper drone presented in this paper tackles these challenges through three main innovations. First, a lightweight, vision-based system for autonomous perch detection using onboard flight hardware detects (imperfect) cylindrical objects found in both natural and artificial environments. Second, an onboard flight planning algorithm autonomously handles the detection, approach and perching flight phases, removing the need for a pilot. Third, a completely passive gripper utilises bistable shell structures to allow for perching on general long narrow features without any precise control inputs or power consumption. This design was successfully validated through both simulation and multiple indoor flights to result in reliable autonomous quadrotor perching in real-world environments.},

note = {Hamaza, S. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:98b694a1-72fc-4a4f-907b-d511cc7f9bdd,

title = {Direct Learning of Home Vector Direction: Incited by Existing Insect-Inspired Approaches for Local Navigation and Wayfinding},

author = {Michiel Firlefyn},

url = {http://resolver.tudelft.nl/uuid:98b694a1-72fc-4a4f-907b-d511cc7f9bdd},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Insects have long been recognized for their ability to navigate and return home using visual cues from their nest’s environment. However, the precise mechanism underlying this remarkable homing skill remains a subject of ongoing investigation. Drawing inspiration from the learning flights of honey bees and wasps, we propose a robot navigation method based on directly learning the home vector directions from visual percepts during the learning flight. Subsequently, the robot will travel away from the nest, come back by odometric means, and eliminate the resultant drift by inferring the home vector orientation from the currently experienced view. In this study, a convolutional neural network is employed as learning mechanism in both simulated and real forest environments. Additionally, a comprehensive performance analysis reveals that the network’s homing abilities closely resemble those observed in real insects, all while only utilizing visual and odometric senses. If all images contain sufficient texture and illumination, the average errors<br/>of the inferred home vectors remain below 24°. Moreover, our investigation reveals a noteworthy insight: the trajectory followed during the initial learning flight, for sample image acquisition, exerts a pronounced impact on the network’s output. For instance, a higher density of sample points in proximity to the nest results in a more consistent return.},

note = {de Croon, G.C.H.E. (mentor); Hagenaars, J.J. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:ca95c8cb-8df3-4d43-9d17-c7b7f54eb1ea,

title = {Evaluating Dynamic Environment Difficulty for Obstacle Avoidance Benchmarking},

author = {MOJI SHI},

url = {http://resolver.tudelft.nl/uuid:ca95c8cb-8df3-4d43-9d17-c7b7f54eb1ea},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Mechanical, Maritime and Materials Engineering; TU Delft Cognitive Robotics},

abstract = {Dynamic obstacle avoidance remains a crucial research area for autonomous systems, such as Micro Aerial Vehicles (MAVs) and service robots. <br/>Efforts to develop dynamic collision avoidance techniques in unknown environments have proliferated in recent years. While these methods exhibit impressive and reliable performance in simpler environments, their efficacy in more challenging settings remains an area ripe for enhancement. The difficulty of these environments arises from a multitude of factors, and currently, no standardized approach exists to quantify this complexity. Additionally, to fairly compare different dynamic collision avoidance strategies, it's essential to assess them in environments with a similar degree of difficulty. Therefore, devising a metric capable of accurately gauging the intricacy of dynamic environments becomes imperative.<br/><br/><br/>Building on this context, this master's thesis endeavors to fill this critical gap through three contributions: 1) The establishment and validation of map difficulty metrics that represent the difficulty of dynamic environments, 2) The introduction of a robust benchmarking pipeline to critically validate the representativeness of the proposed metrics and evaluate various collision avoidance strategies, and 3) The provision of a framework for comparative analysis of different planning strategies, utilizing the introduced map difficulty metric. <br/><br/>The proposed survivability metric effectively captures environmental complexity. Its validity is evidenced by a notable correlation with the success rates of typical collision avoidance methods, with over 1.7 million collision avoidance trials on over six hundred maps, securing a Spearman's Rank correlation coefficient (SRCC) of over 0.9. This metric serves as an indispensable tool for facilitating fair comparisons in this dynamic research domain. More importantly, it offers valuable insights for the future refinement and improvement of dynamic collision avoidance strategies, making a contribution to the continuous advancement of autonomous systems.},

note = {Alonso Mora, J. (mentor); Chen, G. (mentor); Wisse, M. (graduation committee); Hamaza, S. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:bcac496c-6757-4067-b1dd-5d8356486bf8,

title = {EV-LayerSegNet: Self-supervised Motion Segmentation using Event-based Cameras},

author = {Youssef Farah},

url = {http://resolver.tudelft.nl/uuid:bcac496c-6757-4067-b1dd-5d8356486bf8},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Event cameras are novel bio-inspired sensors that capture motion dynamics with much higher temporal resolution than traditional cameras, since pixels react asynchronously to brightness changes. They are therefore better suited for tasks involving motion such as motion segmentation. However, training event-based networks still represents a difficult challenge, as obtaining ground truth is very expensive and error-prone. In this article, we introduce EV-LayerSegNet, the first self-supervised CNN for event-based motion segmentation. Inspired by a layered representation of the scene dynamics, we show that it is possible to learn affine optical flow and segmentation masks separately, and use them to deblur the input events. The deblurring quality is then measured and used as self-supervised learning loss.},

note = {de Croon, G.C.H.E. (mentor); Mooij, E. (graduation committee); Ellerbroek, Joost (graduation committee); Paredes Valles, F. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:e52f0ee0-a859-4177-80e4-268dfd65deca,

title = {Memory Mechanisms in Spiking Neural Networks},

author = {Laurens Lammers},

url = {http://resolver.tudelft.nl/uuid:e52f0ee0-a859-4177-80e4-268dfd65deca},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Neuromorphic sensors, like for example event cameras, detect incremental changes in the sensed quantity and communicate these via a stream of events. Desired properties of these signals such as high temporal resolution and asynchrony are not always fully exploited by algorithms that process these signals. Spiking neural networks (SNNs) have emerged as the algorithms that promise to maximally attain these characteristics and are likely the key to achieving a fully neuromorphic computing pipeline. But, this means that if the SNN is to take full advantage, the event stream must be sent directly and unaltered to the SNN, which in turn implies that all temporal integration should occur inside the SNN. Therefore, it is interesting to investigate the mechanisms that achieve this. This thesis does so through evaluating and comparing the performance of different memory mechanisms in SNNs found in the literature, as well as through an in depth analysis of the inner workings of these mechanisms. The mechanisms include spiking neural dynamics (leaks and thresholds), explicit recurrent connections, and propagation delays. We demonstrate our concepts on two small scale generated 1D moving pixel tasks in preliminary experiments first. After that, we extend our research to compare the memory mechanisms on a real-world neuromorphic vision processing task, in which the networks regress angular velocity given event based input. We find that both explicit recurrency and delays improve the prediction accuracy of the SNN, compared to having just spiking neuronal dynamics. Analysis of the inner workings of the networks shows that the threshold and reset mechanism of spiking neurons play an important role in allowing longer neuron timescales (lower membrane leak). Forgetting (at the right time) turns out to play an important role in memory. Additionally, it becomes apparent that optimizing an SNN with explicit recurrent connections or learnable delays does not lead to the formation of robust spiking neuronal dynamics. In fact, spiking neuronal dynamics are largely ignored, as after optimization virtually no input current is integrated onto the membrane potential in these cases. Instead, we consistently find that a recurrent SNN prefers to build a state solely with the explicit recurrent connections, while an SNN with delays prefers to just use the delays. Therefore, our SNNs with explicit recurrent connections and delays are in fact better described as binary activated RNNs and ANNs, respectively.},

note = {Hagenaars, J.J. (mentor); de Croon, G.C.H.E. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:232b5015-df70-424b-91ab-149ed4d8416a,

title = {An Adaptive Neural Network Quadrotor Trajectory Tracking Controller Tolerant to Propeller Damage},

author = {Mauro Villanueva Aguado},

url = {http://resolver.tudelft.nl/uuid:232b5015-df70-424b-91ab-149ed4d8416a},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {class="MsoNormal" style="margin-bottom:0cm;line-height:normal">Executing quadrotor trajectories accurately and therefore safely is a challenging task. State-of-the-art adaptive controllers achieve impressive trajectory tracking results with slight performance degradation in varying winds or payloads, but at the cost of computational complexity. Requiring additional embedded computers onboard, adding weight and requiring power. Given the limited computational resources onboard, a trade-off between accuracy and complexity must be considered. To this end, we implement "Neural-Fly" a lightweight adaptive neural controller to adapt to propeller damage, a common occurrence in real-world flight. The adaptive neural architecture consists of two components: (I) offline learning of a condition invariant representation of the aerodynamic forces through Deep Neural Networks (II) fast online adaptation to the current propeller condition using a composite adaptation law. We deploy this flight controller fully onboard the flight controller of the Parrot Bebop 1,showcasing its computational efficiency. The adaptive neural controller improves tracking performance by ≈60% over the nonlinear baseline, with minimal performance degradation of just ≈20% with increasing propeller damage.},

note = {de Wagter, C. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:6b397485-750f-4e97-99ec-536ae2933d60,

title = {Rigid airborne docking between a fixed-wing UAV and an over-actuated multicopter},

author = {Jonathas Laffita van den Hove d'Ertsenryck},

url = {http://resolver.tudelft.nl/uuid:6b397485-750f-4e97-99ec-536ae2933d60},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Fixed-wing aircraft fly longer, faster, and further than rotorcraft, but cannot take off or land vertically. Hybrid drones combine VTOL with a wing for forward flight, but the hovering system generally makes them less efficient than a pure fixed-wing. We propose an alternative, in which a rotorcraft is used to assist the fixed-wing UAV with the VTOL portions of the flight. This paper takes the first steps towards this alternative by developing and testing an overactuated rotorcraft that can autonomously dock onto a target at fixed-wing velocities. The control system uses Incremental Non-Linear Dynamic Inversion Control (INDI) to achieve linear accelerations with lateral and longitudinal motors, enabling robust horizontal control independent of attitude. A relative guidance algorithm for the docking approach path is presented, along with a vision sensing approach using ArUco markers and IR LEDs. Successful docking and separation were achieved in the wind tunnel at speeds of up to $15$m/s.},

note = {Smeur, E.J.J. (mentor); Remes, B.D.W. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:5135b7b8-3c4c-46cf-a0cc-e2fbf6da5fff,

title = {Neuro-evolution learned neuromorphic control for a vision-based 3D landing},

author = {Erwin Lodder},

url = {http://resolver.tudelft.nl/uuid:5135b7b8-3c4c-46cf-a0cc-e2fbf6da5fff},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

note = {de Croon, G.C.H.E. (mentor); Stroobants, S. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:5d786e19-6871-4478-bda8-43f7cab20633,

title = {Synthetic Air Data System for Pitot Tube Failure Detection on the Variable Skew Quad Plane},

author = {Frédéric Larocque},

url = {http://resolver.tudelft.nl/uuid:5d786e19-6871-4478-bda8-43f7cab20633},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering; TU Delft Control & Simulation},

abstract = {Pitot tube-free airspeed estimation methods exist for fixed-wing and multirotor configurations, but lack direct applicability to hybrid unmanned air vehicles due to their wide flight envelope and changing dynamics during transition. This work proposes a novel synthetic air data system for the Variable Skew Quad Plane (VSQP) hybrid vehicle to allow airspeed estimation from hover to high speed forward flight and provide pitot tube fault detection. An Extended Kalman Filter fuses Global Navigation Satellite System (GNSS) and inertial measurements using model-independent kinematics equations to estimate wind and airspeed without the use of the pitot tube. The filter is augmented by a simplified vehicle force model. Pitot tube fault detection is achieved with a simple thresholding operation on the pitot tube measurement and the airspeed estimation residual. Accurate airspeed estimation was validated with logged test flight data, achieving an overall 1.62 m/s root mean square error. Using the airspeed estimation, quick detection (0.16 s) of a real-life abrupt pitot tube fault was demonstrated. This new airspeed estimation method provides an innovative approach for increasing the fault tolerance of the VSQP and similar quad-plane vehicles.},

note = {Smeur, E.J.J. (mentor); Remes, B.D.W. (graduation committee); De Ponti, T.M.L. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:3fcc6230-86fc-4848-a6ed-16dd46fea640,

title = {Energy poverty, bridging the gap between housing association and tenant: What measures housing associations can take to aid their tenants who are struggling with energy poverty},

author = {Marvin Cairo},

url = {http://resolver.tudelft.nl/uuid:3fcc6230-86fc-4848-a6ed-16dd46fea640},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Architecture and the Built Environment},

abstract = {Due to rising energy prices, an increasing number of households are experiencing difficulties with the affordability of their energy bills. As a result, households are unable to heat or cool their homes, or use electrical appliances as desired. This is known as energy poverty. This research focuses on energy poverty within housing associations. As two-thirds of households experiencing energy poverty live in housing association homes, this research is specifically targeted at housing associations. The research examines the possible gap between what housing associations are doing to combat energy poverty for their tenants, and what tenants would like to see housing associations do for them. Since renovation is simply too expensive and takes several years, it is excluded from consideration. As a result, housing associations will need to take other measures to help their tenants. This research will look at these taken measures and provides recommendations to housing associations to reduce and possibly solve the gap between what they can do and what tenants want to happen. The main question of this thesis is: What can housing associations do to close the gap between them and their tenants in the social housing sector regarding combating energy poverty?<br/>This research will be carried out based on a qualitative study in which literature will be reviewed, and housing associations and tenant organisations will be interviewed. The aim is to identify the gap between what is desired by tenants and capable of housing associations and to draw up recommendations for housing associations to assist their tenants as well as possible. The recommendations of the research indicate that many of the gaps found during the comparison of the focus groups have to do with communication, both improving communication itself, and setting up communication between tenants and the association to reduce energy poverty.<br},

note = {Hoekstra, J.S.C.M. (mentor); Qian, QK (mentor); Croon, T.M. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:ef354713-924e-4907-a44f-95b67efa638e,

title = {Improving DRL Of Vision-Based Navigation By Stereo Image Prediction},

author = {Luc Ridder},

url = {http://resolver.tudelft.nl/uuid:ef354713-924e-4907-a44f-95b67efa638e},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Although deep reinforcement learning (DRL) is a highly promising approach to learning robotic vision-based control, it is plagued by long training times. This report introduces a DRL setup that relies on self-supervised learning for extracting depth information valuable for navigation. Specifically, a literature study is conducted to investigate the effects of learning how to synthesize one view from the other in a stereo-vision setup without relying on any preliminary knowledge of the camera extrinisics and how it can be integrated for its downstream use for an obstacle avoidance task. As such, the literature study concludes that competitive geometry-free monocular-to-stereo image view synthesis is feasible due to recent developments in computer vision. The scientific paper further develops concepts proposed in the literature study and benchmarks the proposed architectures on depth estimation benchmarks for KITTI. Competitive results are achieved for view synthesis and despite sub-optimal performance compared to state-of-the-art monocular depth estimation, an ability to encode depth and detect shapes is present and, therefore, satisfactory for the application to DRL. Additionally, the research examines the benefits of using the latent space of a view synthesis architecture compared to other feature extractor methods as an input to the PPO agent implemented as auxiliary tasks. This method achieves quicker convergence and better performance for an obstacle avoidance task in a simulated indoor environment than the autoencoding feature extractor and end-to-end DRL methods. It is only outperformed by the monocular depth estimation feature extractor method. Overall, this research provides valuable insights for developing more efficient and effective DRL methods for monocular camera-based drones. Finally, the complementary code for this research can be found: urlhttps://github.com/ldenridder/drl-obstacle-avoidance-view-synthesis.},

note = {de Croon, G.C.H.E. (mentor); Wu, Y. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:2557c822-2360-4d2c-a6e9-0e05182c5c15,

title = {Aerobatic maneuvering of Autonomous Hybrid UAVs: Trajectory Tracking using INDI in the Control Frame},

author = {Koen Engelen},

url = {http://resolver.tudelft.nl/uuid:2557c822-2360-4d2c-a6e9-0e05182c5c15},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Unmanned Aerial Vehicles (UAVs) are increasingly being used in various applications, which demand longer endurance, extended range, and high maneuverability. These requirements necessitate the development of effective control methods for Hybrid UAVs. In this paper, we propose an outer loop Incremental Nonlinear Dynamic Inversion (INDI) controller for Hybrid UAVs, based on an analytically derived control effectiveness to control the linear acceleration of the UAV. The control effectiveness is derived in a new frame that does not show singularities, technically allowing controlled flight at all attitudes. For trajectory tracking purposes, a Proportional Derivative (PD) controller is added. In simulation the proposed controller shows comparable results to already existing INDI controllers for hover and forward flight. When performing loop the loops it is shown that the proposed control system is able to handle high roll angles, while the already existing INDI controller crashed.},

note = {Smeur, E.J.J. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:9b27d79d-d876-466d-b980-562c03552e6b,

title = {Autonomous Control for Orographic Soaring of Fixed-Wing UAVs},

author = {Tom Suys},

url = {http://resolver.tudelft.nl/uuid:9b27d79d-d876-466d-b980-562c03552e6b},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Prolonging the endurance of fixed-wing UAVs is crucial for achieving complex missions, yet their limited battery life poses a significant challenge. In response, this research proposes a novel approach to extend the endurance of fixed-wing UAVs by enabling autonomous soaring in an orographic wind field. The goal of our research is to develop a controller that can identify feasible soaring regions and autonomously maintain position control without using any throttle. Soaring flight is desirable as it results in a low energy cost with zero throttle usage. However, without throttle usage, the longitudinal motion of the UAV is an under-actuated system, presenting control challenges. The concept of a target gradient line (TGL) is introduced as part of the control algorithm that addresses these challenges and autonomously finds the equilibrium soaring position where sink rate and updraft are in equilibrium. Experimental tests showed promising results, demonstrating the controller’s effectiveness in maintaining autonomous soaring flight in a non-static wind field. We also demonstrate a single degree of control freedom in the soaring position through manipulation of the TGL.},

note = {de Croon, G.C.H.E. (mentor); Remes, B.D.W. (mentor); Hwang, S. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:fd5e484b-bdd4-42e7-8cdc-70de94462858,

title = {A Centralised Approach to Aerial Manipulation on Overhanging Surfaces},

author = {Martijn Brummelhuis},

url = {http://resolver.tudelft.nl/uuid:fd5e484b-bdd4-42e7-8cdc-70de94462858},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Aerial physical interaction opens the door for many operations at height to be automatised using aerial robots. This research presents a novel manipulator design mounted on a traditional quadrotor, which utilises both mechanical and software compliance to perform physical interaction on vertical walls and overhanging surfaces, such as those found under bridges. A centralised impedance control scheme allows direct control of the end-effector pose without needing separate modes for free-flight and contact. A spring-loaded prismatic joint provides passive compliance while doubling as a force-feedback for the impedance controller through measuring the spring displacement. Simulation and flight experiments prove the feasibility and robustness of this approach for exchanging high forces at height, with a total of 44 successful experiments carried out in four sets. An average maximum force of 5.66 N or 19.3% of the system's weight was achieved over one set of 11 experiments.},

note = {Hamaza, S. (mentor); Smeur, E.J.J. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:efa9ee47-b200-4a52-b61d-d2c8e5b6fb78,

title = {Guidance & Control Networks for Time-Optimal Quadcopter Flight},

author = {Sebastien Origer},

url = {http://resolver.tudelft.nl/uuid:efa9ee47-b200-4a52-b61d-d2c8e5b6fb78},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Reaching fast and autonomous flight requires computationally efficient and robust algorithms. To this end, we train Guidance & Control Networks to approximate optimal control policies ranging from energy-optimal to time-optimal flight. We show that the policies become more difficult to learn the closer we get to the time-optimal ’bang-bang’ control profile. We also assess the importance of knowing the maximum angular rotor velocity of the quadcopter and show that over- or underestimating this limit leads to less robust flight. We propose an algorithm to identify the current maximum angular rotor velocity onboard and a network that adapts its policy based on the identified limit. Finally, we extend previous work on Guidance & Control Networks by learning to take consecutive waypoints into account. We fly a 4×3m track in similar lap times as the differential-flatness-based minimum snap benchmark controller while benefiting from the flexibility that Guidance & Control Networks offer.},

note = {de Wagter, C. (mentor); de Croon, G.C.H.E. (mentor); Izzo, Dario (mentor); Ferede, R. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:f7ec9e1c-15db-4982-b2fa-4ba0f51a5b91,

title = {Aerial Perching via Active Touch: Embodying Robust Tactile Grasping on Aerial Robots},

author = {Anish Jadoenathmisier},

url = {http://resolver.tudelft.nl/uuid:f7ec9e1c-15db-4982-b2fa-4ba0f51a5b91},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Aerial manipulators, characterized by their ability to actively engage with the environment, are gaining popularity for their versatility in performing diverse tasks.<br/>This research focuses on augmenting the capabilities of aerial manipulators through the integration of tactile feedback, specifically employing a compliant bio-inspired three-fingered manipulator equipped with tactile capacitive sensors on each finger. The manipulator is affixed to a drone, enabling tactile-guided navigation for precise object localization, subsequent grasping, and perching. Additionally, a grasp evaluator assesses grasp quality, allowing the system to adapt by suggesting alternative grasp locations after an initial attempt is unsuccessful. A comparative analysis between the system’s performance using tactile feedback and open-loop perching/grasping in perching scenarios demonstrates that the grasp evaluator improves the perching success rate by 55%-point and increases the allowable object uncertainty by 0.14 [m]. These findings highlight the efficacy of this approach in advancing aerial manipulator capabilities.},

note = {Hamaza, S. (mentor); de Croon, G.C.H.E. (graduation committee); Pool, D.M. (graduation committee); Bredenbeck, A. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:bfad7b0e-f6db-49d0-8642-5e4fbc6e3861,

title = {Collaborative Payload Carrying with Multiple MAVs},

author = {Huamin Liu},

url = {http://resolver.tudelft.nl/uuid:bfad7b0e-f6db-49d0-8642-5e4fbc6e3861},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {The transportation of payloads utilizing multiple drones presents a promising application for lifting heavier loads that exceed the payload capacity of a single drone. However, the cable-suspended payload introduces significant challenges to the system, and this research area remains relatively unexplored. In this work, a novel solution for payload-carrying application is proposed. First, the dynamics of cable-suspended payload transportation using multiple quadrotors, taking into account the influence of drag forces on the quadrotors are studied. A nonlinear optimization is employed to control the payload while distributing the control effort required for manipulating the suspended load over the drones in the formation while ensuring both tension constraints and collision avoidance between drones in the formation. The feasible path commands for formation agents are computed from the optimization. One of the critical aspects for controlling such a system is the load-introduced force, which exhibits rapid and complex variations. To address this, an extended state observer is employed to estimate the load force, eliminating the need for a tension sensor. In pursuit of a robust framework, a formation reset strategy is also developed, allowing to maintain load tracking performance and ensure the safety of formation agents, even in the event of a malfunction in one of the drones. A series of simulations are conducted to validate the effectiveness and robustness against disturbance and suspension failure of the proposed strategy and controllers. Results demonstrate that the whole multi-lift system can handle external disturbances, model uncertainties regarding drone inertia, mass and load mass, as well as suspension failures.},

note = {Smeur, E.J.J. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:34213dbf-32ad-4f8d-b0f0-ed398608d682,

title = {Modular Neural Network Navigation for Autonomous Nano Drone Racing},

author = {Federico Magri},

url = {http://resolver.tudelft.nl/uuid:34213dbf-32ad-4f8d-b0f0-ed398608d682},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {In this study, we present a first step towards a cutting-edge software framework that will enable autonomous racing capabilities for nano drones. Through the integration of neural networks tailored for real-time operation on resource-constrained devices. A lightweight Convolutional Neural Network, with the Gatenet architecture, is adjusted for reduced computational demand and is successfully deployed on a GAP8 processor at a rate of 16$Hz$. This network provides gates' size and location data for the subsequent positioning algorithm. A second neural network, trained through reinforcement learning, governs the drone's guidance and control systems, demonstrating a remarkable rate of 167$Hz$ on an STM32F405 processor. The attitude rates and thrust outputted by this network are then fed to an attitude rate PID controller.<br/><br/>The research shows that state-of-the-art neural networks for drone racing can be deployed on nano drones, despite their limited processing power. Nonetheless, the study demonstrated specific limitations, such as the perception network's sensitivity to white pixels in the image reducing its effectiveness when light sources are present in the scene. These findings underscore the importance of dataset composition and the need for diverse training scenarios to enhance the neural network's generalizability and performance in real-world applications.},

note = {de Wagter, C. (mentor); de Croon, G.C.H.E. (mentor); Ferede, R. (mentor); Bahnam, S.A. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:c56b290f-8c43-4649-8360-13b7652710e8,

title = {Tilt-rotor Tailsitter Global Acceleration Control: Behavioural Cloning of a Nonlinear Model Predictive Controller},

author = {Alexis Wissen},

url = {http://resolver.tudelft.nl/uuid:c56b290f-8c43-4649-8360-13b7652710e8},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Capable of both vertical take-off and landing and forward flight, tail-sitters are a versatile class of UAVs with a large range of potential applications. A variant of tailsitters using tilt-rotors instead of ailerons for pitch and roll control has been proposed to mitigate the reduced control authority at low to zero velocities. The control of the translational dynamics for this platform is uniquely challenging. The extended flight envelope requires the controller to be able to perform hover and forward flight which are two flight phases with very different dynamics. Additionally, the tilt-rotor mechanism used to control the system is highly nonlinear which adds to the challenge. This paper presents a novel acceleration controller using Nonlinear Model Predictive Control (NMPC) in addition to the use of behavioural cloning to mimic the NMPC using a feedforward neural network. It is shown that behavioural cloning does successfully transfer general flight characteristics but that the performance is degraded with respect to the NMPC. Additionally, a sensitivity analysis was performed to investigate the effects of improper parameter estimation on controller performance. The most interesting result from this analysis is the strong sensitivity of both controllers to changes in centre of gravity location and mass.},

note = {Smeur, E.J.J. (mentor); Ma, Z. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:bd663e11-947f-4ae1-ae2d-860d44fba7af,

title = {Factor graphs for inventory label scanning in warehouse},

author = {DARSHAN KALYANASUNDARAM},

url = {http://resolver.tudelft.nl/uuid:bd663e11-947f-4ae1-ae2d-860d44fba7af},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Mechanical, Maritime and Materials Engineering},

abstract = {A key important part of a warehouse operation is to keep track of the products in the warehouse. Traditionally, handheld scanners are used to scan the products to perform a stock count. The advancements in robotics have paved the way for new technologies that can improve the scanning process. This work shows how prior knowledge and warehouse structure can be used to perform scanning operations. The method uses a localized camera in the warehouse whose estimates drift over time and the knowledge about the environment to estimate the correct location of the product using factor graphs. The proposed method shows by exploiting the structure of the warehouse the drift in the VIO(Visual Inertial Odometry) can be reduced and the position estimation of the location labels can be improved},

note = {de Croon, G.C.H.E. (mentor); Alonso Mora, J. (mentor); Ozo, Michaël (mentor); Caesar, H.C. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:ab98e53a-5931-4b78-98ec-f0cb6df20986,

title = {GSL-Bench: High Fidelity Gas Source Localization Benchmarking},

author = {Hajo Erwich},

url = {http://resolver.tudelft.nl/uuid:ab98e53a-5931-4b78-98ec-f0cb6df20986},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Gas Source Localization (GSL) is a challenging field of research within the robotics community. Existing methods vary widely and each has its own strengths and weaknesses. Existing GSL evaluations vary in environment size, wind conditions, and gas simulation fidelity, thereby complicating objective comparison between algorithms. They also lack photo-realistic rendering for the integration of obstacle avoidance. In this paper, we propose GSL-Bench, a benchmarking suite to evaluate the performance of GSL algorithms. GSL-Bench features high-fidelity graphics and gas simulation. Realism is further increased by simulating relevant gas and wind sensors. Scene generation is simplified with the introduction of AutoGDM+, capable of procedural environment generation, CFD and particle-based gas dispersion simulation. To illustrate GSL-Bench's capabilities, three algorithms are compared in six warehouse settings of increasing complexity: E. Coli, dung beetle and a random walker. Our results demonstrate GSL-Bench's ability to provide valuable insights into algorithm performance.},

note = {de Croon, G.C.H.E. (mentor); Duisterhof, B.P. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:6320374a-b84d-4bbf-be48-10cee914b9e0,

title = {Application of Control Barrier Functions to Collision Free Model Predictive Control: Robust UAV Trajectories with MPC-CBF and Euclidean Signed Distance Fields},

author = {Rinto Vries},

url = {http://resolver.tudelft.nl/uuid:6320374a-b84d-4bbf-be48-10cee914b9e0},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Recent literature in real-time trajectory planning has proposed using Control Barrier Functions (CBFs) as collision constraints in Model Predictive Control (MPC) for efficient guidance, a concept referred to as MPC-CBF. This concept has been explored for both first and second-order CBFs. However, these approaches relied on an analytical description of the environment. Building upon this, we propose combining MPC-CBF with Euclidean Signed Distance Fields (ESDFs), eliminating the need for such an analytical model of the environment. Notably, we extend this approach to a new field by applying it to Unmanned Aerial Vehicles (UAVs). Through simulations, we compare flown trajectories and noise robustness for distance constraints, first-order CBF constraints and second-order CBF constraints. First-order CBF constraints outperform distance constraints, excelling in path planning and noise resilience. Second-order CBF constraints face challenges due to numerical approximations of the hessian of the ESDF and stricter dependency on an accurate acceleration model, limiting their practicality for UAVs. The proposed control framework was tested by safely maneuvering an enterprise inspection drone around a Boeing 787-9 aircraft inside an aircraft hangar, confirming its effectiveness in collision avoidance and real-world scenarios.},

note = {Smeur, E.J.J. (mentor); Horstink, Thomas (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:319f4f93-0590-4f9e-8029-2911f61db477,

title = {The development of a distributed electric propulsion (DEP) noise model},

author = {Chris Bononi Bello},

url = {http://resolver.tudelft.nl/uuid:319f4f93-0590-4f9e-8029-2911f61db477},

year  = {2023},

date = {2023-01-01},

school = {TU Delft Aerospace Engineering; TU Delft Aircraft Noise and Climate Effects},

note = {Snellen, M. (mentor); Yin, F. (graduation committee); Smeur, E.J.J. (graduation committee); Heblij, S. (mentor); de Haan, W (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:4e100997-a5b3-4863-a312-4721296fcdba,

title = {Towards landing a deep-stalled flying-wing in a powered flat spin: a proof of concept},

author = {Matteo Barbera},

url = {http://resolver.tudelft.nl/uuid:4e100997-a5b3-4863-a312-4721296fcdba},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Flying-wings show great potential for a vast number of applications, in both commercial and military sectors, thanks to their long range and fast forward flight, but suffer due to their lack of vertical take-off and landing capabilities. This paper presents a proof of concept for a novel landing method for a conventional flying wing that does not introduce additional weight dedicated only to the landing phase, with the aim of controlling a deep-stalled flying-wing in a powered flat spin. Through cyclic actuation of the servo motors and elevons, lateral forces as well as moments can be generated to control the position and attitude of the rotation plane. A successful indoor experiment was performed with a modified Parrot Disco in a controlled environment. Outdoor tests, however, failed to replicate the indoor results due to additional challenges present in the real flight conditions. A number of key challenges were identified, and the insights gained in this research lay an initial foundation for future work on this topic.},

note = {de Wagter, C. (mentor); Remes, B.D.W. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:1136170f-3c4b-43b8-8b43-09e1e52d3bfd,

title = {Cooperative Relative Localization in MAV Swarms with Ultra-wideband Ranging},

author = {Changrui Liu},

url = {http://resolver.tudelft.nl/uuid:1136170f-3c4b-43b8-8b43-09e1e52d3bfd},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Relative localization (RL) is essential for the successful operation of micro air vehicle (MAV) swarms. Achieving accurate 3-D RL in infrastructure-free and GPS-denied environments with only distance information is a challenging problem that has not been satisfactorily solved. In this work, based on the range-based peer-to-peer RL using the ultra-wideband (UWB) ranging technique, we develop a novel UWB-based cooperative relative localization (CRL) solution which integrates the relative motion dynamics of each host-neighbor pair to build a unified dynamic model and takes the distances between the neighbors as bonus information. Observability analysis using differential geometry shows that the proposed CRL scheme can expand the observable subspace compared to other alternatives using only direct distances between the host agent and its neighbors. In addition, we apply the kernel-induced extended Kalman filter (EKF) to the CRL state estimation problem with the novel-designed Logarithmic-Versoria (LV) kernel to tackle heavy-tailed UWB noise. Sufficient conditions for the convergence of the fixed-point iteration involved in the estimation algorithm are also derived. Comparative Monte Carlo simulations demonstrate that the proposed CRL scheme combined with the LV-kernel EKF significantly improves the estimation accuracy owing to its robustness against both the measurement outliers and incorrect measurement covariance matrix initialization. Moreover, with the LV kernel, the estimation is still satisfactory when performing the fixed-point iteration only once for reduced computational complexity.},

note = {de Croon, G.C.H.E. (mentor); Pfeiffer, S.U. (mentor); Mazo, M. (graduation committee); de Wagter, C. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:baf5b7df-0e0f-45da-8b70-c7c95ead79b6,

title = {Attitude Control of a Tilt-rotor Tailsitter Micro Air Vehicle Using Incremental Control},

author = {Gervase Lovell-Prescod},

url = {http://resolver.tudelft.nl/uuid:baf5b7df-0e0f-45da-8b70-c7c95ead79b6},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {By combining the ability to hover with a wing for fast and efficient horizontal flight, hybrid unmanned aircraft extend the flight envelope and therefore mission capabilities of unmanned aircraft. However, this comes at a cost: increased complexity control-wise and being more susceptible to wind disturbances. This susceptibility to wind gusts is particularly problematic for tailsitters as during hovering and vertical flight their wing is perpendicular to horizontal wind disturbances, often leading to actuator saturation. This paper presents a novel tailsitter micro air vehicle with two leading edge tilting rotors serving as its only actuators. It is shown that thrust vectoring generates sufficient control moment generation alleviating actuator saturation. Incremental nonlinear dynamic inversion (INDI) is implemented for attitude control and is demonstrated to compensate for unmodeled forces and moments whilst only relying on actuator control effectiveness and knowledge of actuator dynamics.},

note = {Smeur, E.J.J. (mentor); Ma, Z. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:1caff7b3-5c17-4b80-abea-19c629ce6051,

title = {Verification & Validation of Focus of Expansion estimation algorithm employing event-based optic flow},

author = {Stefan Knoops},

url = {http://resolver.tudelft.nl/uuid:1caff7b3-5c17-4b80-abea-19c629ce6051},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Event based vision has recently attracted a lot of attention. High data rates and robustness to lighting variations make it a valid option for indoor navigation. The previously developed FAITH algorithm calculates a possible Focus of Expansion<br/>area based on negative half-planes generated by optic flow and by employing a RANSAC search, a fast and reliable Focus of Expansion estimation can be performed. This paper builds upon this algorithm by verifying and validating the<br/>algorithm, improving the derotation capabilities and optimising for computational efficiency. Compared to earlier work, a higher accuracy and an increased robustness are realised by improving the data handling. Simulator results show accuracies in the range of 2 to 5 degrees. Online testing on a drone shows accuracies of up to 5 degrees while obtaining calculation times of only<br/>2 · 10−3s and rates of 140Hz. Comparing the method to an alternative shows higher accuracy and better suitability to normal flow. Further research may contribute to more stable results and explore different hardware solutions.},

note = {de Croon, G.C.H.E. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:574db806-6096-4600-9926-3d737d1ee7da,

title = {NanoFlowNet: Real-time optical flow estimation on a nano quadcopter},

author = {Rik Bouwmeester},

url = {http://resolver.tudelft.nl/uuid:574db806-6096-4600-9926-3d737d1ee7da},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Nano quadcopters are small, agile, and cheap platforms well suited for deployment in narrow, cluttered environments. Due to their limited payload, nano quadcopters are highly constrained in processing power, rendering conventional vision-based methods for autonomous navigation incompatible. Recent machine learning developments promise high-performance perception at low latency, while novel ultra-low power microcontrollers augment the visual processing power of nano quadcopters. In this work, we present NanoFlowNet, an optical flow CNN that, based on the semantic segmentation architecture STDC-Seg, achieves real-time dense optical flow estimation on edge hardware. We use motion boundary ground truth to guide the learning of optical flow, improving performance with zero impact on latency. Validation on MPI-Sintel shows the high performance of the proposed method given its constrained architecture. We implement the CNN on the ultra-low power GAP8 microcontroller and demonstrate it in an obstacle avoidance application on a 34 g Bitcraze Crazyflie nano quadcopter.},

note = {de Croon, G.C.H.E. (mentor); Paredes-Vallés, Federico (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:4e4e333d-643f-43b9-99cb-650d697f5baa,

title = {ADAPT: A 3 Degrees of Freedom Reconfigurable Force Balanced Parallel Manipulator for Aerial Applications},

author = {KARTIK SURYAVANSHI},

url = {http://resolver.tudelft.nl/uuid:4e4e333d-643f-43b9-99cb-650d697f5baa},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Mechanical, Maritime and Materials Engineering},

abstract = {In this work, we present the ADAPT, a novel reconfigurable force-balanced parallel manipulator with pantograph legs for spatial motions applied underneath a drone. The reconfigurable aspect allows different motion-based 3-DoF operation modes like translational, rotational, mixed, planar without disassembly. For the purpose of this study, the manipulator is used in translation mode only. A kinematic model is developed and validated for the manipulator. The design and motion capabilities are also validated both by conducting dynamics simulations of a simplified model on MSC ADAMS, and experiments on the physical setup.<br/>The force-balanced nature of this novel design decouples the motion of the manipulator’s end-effector from the base, zeroing the reaction forces, making this design ideally suited for aerial manipulation in unmanned aerial vehicles (UAVs) applications, or generic floating-base applications.},

note = {van der Wijk, V. (mentor); Hamaza, S. (mentor); Herder, J.L. (graduation committee); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}

@mastersthesis{uuid:66c34a84-5b47-49dd-b560-2836d9696e3c,

title = {Incremental Nonlinear Dynamic Inversion controller - structural vibration coupling: Study of the phenomenon and the existing solutions},

author = {Alessandro Collicelli},

url = {http://resolver.tudelft.nl/uuid:66c34a84-5b47-49dd-b560-2836d9696e3c},

year  = {2022},

date = {2022-01-01},

school = {TU Delft Aerospace Engineering},

abstract = {Incremental Nonlinear Dynamics Inversion (INDI) flight controllers are sensor-based control systems, that are robust towards model uncertainty and with good disturbance rejection characteristics. These controllers show coupling effects in structural modes when implemented in specific flying vehicles with low-frequency structural motions. This paper investigates different INDI implementations, standard INDI, hybrid INDI, and notch filter placement in the INDI loop via simulation and flight tests on the Nederdrone. System identification of the structural characteristics of the vehicle and the system’s yaw dynamics are executed via ground vibration and hover flight tests. Closed-loop behaviour of theINDI inner-loop, disturbance rejection performance, and outer loop step-tracking performance was assessed with dedicated flight tests. The investigated INDI solutions show similar disturbance rejection and outer-loop step tracking performance, while the hybrid INDI solution performs a better nonlinear dynamic inversion. <br/>Index Terms—INDI, complementary filter, unmanned vehicle, flight control system structural motion coupling},

note = {Smeur, E.J.J. (mentor); Pollack, T.S.C. (mentor); Delft University of Technology (degree granting institution)},

keywords = {},

pubstate = {published},

tppubtype = {mastersthesis}

}