Unmanned Surface Vessels (USVs) face significant control challenges due to uncertain environmental disturbances, such as waves and currents. This thesis proposes a trajectory tracking controller based on Active Disturbance Rejection Control (ADRC) implemented on the DUS V2500, a seagoing USV developed by Demcon Unmanned Systems. A custom simulation incorporating realistic wave, wind, and current disturbances is developed to validate the performance of the controller. Simulated experiments are supported by further validation through field tests in the harbour of Scheveningen, the Netherlands, and at sea. Simulation results demonstrate that ADRC significantly reduces cross-track error across all tested conditions compared to a baseline PID controller, but increases control effort and energy consumption. Field trials in Scheveningen, the Netherlands, confirm a reduction in cross-track error, while revealing a further increase in energy consumption during sea trials compared to the baseline controller. Modifications to the controller are proposed to improve its efficiency, and the impacts on tracking performance are discussed.

To safely and efficiently solve motion planning problems in multi-agent settings, most approaches attempt to solve a joint optimization that explicitly accounts for the responses triggered in other
agents. This often results in solutions with an exponential computational complexity, making these methods intractable for complex scenarios with many agents. While sequential predict-and-plan approaches are more scalable, they tend to perform poorly in highly interactive environments. This paper proposes a method to improve the interactive capabilities of sequential predict-and-plan methods in multi-agent navigation problems by introducing predictability as an optimization objective. We interpret predictability through the use of general prediction models, by allowing agents to predict themselves and estimate how they align with these external predictions. We formally introduce this behavior through the free-energy of the system, which reduces (under appropriate bounds) to the Kullback-Leibler divergence between plan and prediction, and use this as a penalty for unpredictable trajectories. The proposed interpretation of predictability allows agents to more robustly leverage prediction models, and fosters a ‘soft social convention' that accelerates agreement on coordination strategies without the need of explicit high level control or communication. We show how this predictability-aware planning leads to lower-cost trajectories and reduces planning effort in a set of multi-robot problems, including autonomous driving experiments with human driver data, where we show that the benefits of considering predictability apply even when only the ego-agent uses this strategy.

Mobile manipulators, which integrate a robotic arm on a mobile base, are increasingly being explored and deployed in sectors such as healthcare, logistics, and aerospace. While motion planning for these systems has been studied in single-agent scenarios, the use of multiple robots to enhance efficiency and accelerate task completion in multi-agent settings remains largely unexplored, particularly in real-world environments. Extending motion planning to multi-mobile manipulators introduces challenges in real-time performance, collision avoidance, and coordination. To address these, this thesis proposes a decentralized Model Predictive Control (MPC) framework with a double integrator as dynamic model, denoted as MPC-d, tailored for multi-mobile manipulators operating in shared workspaces. It integrates optimization-based planning with robust state estimation, ensuring effective collision avoidance. Furthermore, a prioritized heuristic is introduced, leveraging the prediction horizon of MPC to resolve potential livelocks. The framework is validated through simulations and real-world experiments. Simulations compare MPC-d with MPC using a triple-integrator model (MPC-t) and a state-of-the-art geometric planner, called Geometric Fabrics (GF). Results demonstrate that MPC-d achieves comparable task success rates and collision avoidance compared to GF in pick-and-place scenarios while requiring less computation time than MPC-t. Real-world experiments confirm the framework’s viability, showcasing effective collision avoidance, enhanced efficiency from the prioritized heuristic, and consistency with simulation outcomes. Although MPC-d incurs higher computational costs than reactive geometric methods, it provides reliable performance and motion prediction of other agents in multi-agent settings.

The application of multi-robot systems has gained popularity in recent years. Multi-robot systems show great potential in scaling up robotic applications in surveillance, monitoring, and exploration. Although single robots can already be used to automatize search and rescue, and surveillance tasks, their capability to solve their given task is still dependent on the region of interest size. For example, in expansive environments, a single robot’s ability to cover or surveil the entire area effectively diminishes, resulting in information gaps. To this end, this thesis aims to improve multi-robot coordination and navigation in active perception tasks, e.g. exploration and surveillance. The multi-robot coordination problem aims to determine when and with whom multiple robots should exchange information to avoid collisions while navigating to their goal. The active perception problem involves guiding a sensing robot to positions rich in task-relevant information. The primary objective of this dissertation is to provide algorithmic contributions, specifically focusing on the obtention of robust policies that can adapt and scale to fleets and tasks of varying sizes. To this end, this thesis leverages the combination of high-level policies, learned through Reinforcement Learning, with robust low-level optimal control policies. Firstly, the communication-based coordination problem for decentralized multi-robot systems is formally defined and modeled as a Markov Decision Process. This thesis then proposes a novel communication policy for decentralized multi-robot systems, improving coordination and collision avoidance. The proposed policy, learned via Reinforcement Learning, allows robots to selectively communicate, requesting trajectory plans from potential risks while assuming constant velocities for others. This utilizes an attention-based neural network architecture for scalability, integrated with Non-Linear Model Predictive Control for safe and robust motion planning. When tested with 12 robots, it reduced communications compared to alternatives, maintaining safety. Scalable and robust, it performed well with different team sizes and in the presence of observation noise. Real-world tests on quadrotors confirmed its practical applicability. The Active Perception problem represents the second major challenge addressed in this dissertation. Specifically, this thesis addresses the challenge of using a drone to collect semantic information for classifying multiple moving targets, focusing on computing control inputs for optimal viewpoints. This task is complicated by the variable amount of targets to classify in the region of interest, and the use of a ”black-box" classifier, like a deep learning neural network, which lacks clear analytical relationships between viewpoints and outputs. This thesis proposes an attention-based architecture, trained with Reinforcement Learning (RL), which determines the best viewpoints for the drone to gather evidence from multiple unclassified targets, considering their movement, orientation, and potential occlusions. A low-level MPC controller then guides the drone to these viewpoints. The approach outperforms various baselines and shows adaptability to new scenarios and scalability to numerous targets with varying movement dynamics. To conclude the thesis, the previous approach is extended to more realistic and larger environments where targets need to be localized, tracked, and then classified. This thesis introduces a novel decentralized hybrid multi-camera system designed for surveillance and monitoring applications. Traditional fixed camera networks suffer from blind spots and backlighting issues. A decentralized hybrid framework is proposed that integrates both static and mobile cameras to actively and dynamically enhance critical information gathering. All networked cameras collaborate to monitor and localize people in the environment by comparing their local information. The mobile camera is guided by a viewpoint control policy to maximize semantic information from observed targets. The framework was implemented in a photorealistic environment using Unreal Engine and enabled distributed communications through the Robot Operating System (ROS), bridging the gap between simulation and real-world applications. Results in large environments demonstrate the advantages of collaborative mobile cameras over static and individual setups both in target identification and tracking accuracy, respectively. In crowded scenarios, mobile cameras excel in avoiding occlusions and capturing desired viewpoints, improving the percentage of classified tracked targets compared to static setups. Qualitatively, mobile cameras provide superior target observation quality unmatched by the static framework. In summary, this thesis makes significant contributions that are validated through extensive evaluations in simulated photo-realistic environments and with commercial drones, demonstrating the potential for practical applications. Despite the progress, the thesis acknowledges the remaining challenges in deploying multi-robot systems in realworld perception tasks, especially when the policy is learned, and suggests directions for future research. @en

Autonomous motion planning requires the ability to safely reason about learned trajectory predictors, particularly in settings where an agent can influence other agents' behavior. These learned predictors are essential for anticipating the future states of uncontrollable agents, whose decision-making process can be difficult to model analytically. Thus, uncertainty quantification of these predictors is crucial for ensuring safe planning and control. In this work, we introduce a framework for interactive motion planning in unknown dynamic environments with probabilistic safety assurances. We adapt a model predictive controller (MPC) to distribution shifts in learned trajectory predictors when other agents react to the ego agent's plan. Our approach leverages tools from conformal prediction (CP) to detect when the other agent's behavior deviates from the training distribution and employs robust CP to quantify the uncertainty in trajectory predictions during these agent interactions. We propose a method for estimating interaction-induced distribution shifts during runtime and the Huber quantile for enhanced outlier detection. Using a KL divergence ambiguity set that upper bounds the distribution shift, our method constructs prediction regions with probabilistic assurances in the presence of distribution shifts caused by interactions with the ego agent. We evaluate our framework in interactive scenarios involving navigation around autonomous vehicles in the BITS simulator, demonstrating enhanced safety and reduced conservatism.

Autonomous target search is crucial for deploying Micro Aerial Vehicles (MAVs) in emergency response and rescue missions. Existing approaches either focus on 2D semantic navigation in structured environments – which is less effective in complex 3D settings, or on robotic exploration in cluttered spaces – which often lacks the semantic reasoning needed for efficient target search. This thesis overcomes these limitations by proposing a novel framework that utilizes semantic reasoning to minimize target search and exploration time in unstructured environments using a MAV. Specifically, the open vocabulary inference capabilities of Large Language Models are employed to embed semantic relationships in segmentation images. An active perception pipeline is then developed to guide exploration toward semantically relevant regions of 3D space by biasing frontiers and selecting informative viewpoints. Finally, a combinatorial optimization problem is solved using these viewpoints to create a plan that balances information gain with time costs, facilitating rapid location of the target. Evaluations in complex simulation environments show that the proposed method consistently outperforms baselines by quickly finding the target while maintaining reasonable exploration times. Real-world experiments with a MAV further demonstrate the method's ability to handle practical constraints like limited battery life, small sensor range, and semantic uncertainty.

Human Mesh Recovery (HMR) frameworks predict a comprehensive 3D mesh of an observed human based on sensor measurements. The majority of these frameworks are purely image-based. Despite the richness of this data, image-based HMR frameworks are vulnerable to depth ambiguity, resulting in mesh inaccuracies in 3D space. Several HMR frameworks in the literature use LiDAR data to avoid this depth ambiguity. However, these frameworks are either only LiDAR-based, which limits performance due to LiDAR sparseness and limited training data, or they are model-free, making the resulting meshes vulnerable to artifacts and limited detail. Therefore, this work introduces SMPLify-3D, an optimization-based HMR framework that combines the richness of image data and the depth information within sparse LiDAR data to improve the 3D mesh inaccuracies of image-based HMR frameworks. The proposed framework consists of three main steps: 1) a body part visibility filter based on the 2D detected keypoints, 2) rough alignment between the mesh and the observed LiDAR point cloud using the ICP algorithm, and 3) an optimization scheme, inspired by SMPLify, that modifies the actual pose and shape of the mesh to improve both the 3D and image alignment. SMPLify-3D is versatile compared to other methods and outperforms image-based and SMPL-compatible LiDAR-based HMR frameworks by improving the Per-Vertex-Error (PVE) with 45% and 26% on the 3DPW and HumanM3 datasets respectively. Multiple quantitative experiments are conducted to show the effects of LiDAR noise and sparsity on the framework’s performance. Additionally, qualitative results illustrate how the proposed method achieves superior results on out-of-sample data recorded by a mobile robot. The source code for this work is available at: https://github.com/guidodumont/SMPLify-3D

Brachytherapy (BT) is an essential component in the curative treatment of cervical cancer. With commercial BT implant devices, called applicators, the radioactive sources can only be positioned in fixed intracavitary channels or in a fixed array of interstitial needles. Patient-tailored BT applicators containing individualised needle channels have the potential to enhance the delivered dose to the tumour while minimising tissue damage of the organs-at-risk (OARs). During the optimisation process of the needle channels, several constraints must be considered. The needle paths in the applicator are curvature-constrained, and the needles should not collide with each other or potentially perforate OARs. Furthermore, clinicians can impose additional constraints based on their preferences.

Existing approaches in literature for optimising needle placement in patient-tailored BT applicators for cervical cancer treatment have employed various algorithms. However, these approaches often rely on significant approximations. For instance, some assume that all needles are inserted in parallel or focus solely on optimising geometric coverage as opposed to optimising the dose distribution. Moreover, the few methods in literature incorporating needle path planning in the optimisation process require manually pre-specified dwell positions. To improve dose conformity and minimise the dependence on the clinician’s expertise and time, there is a need for the development of software that optimises the needle placement and paths without resorting to these severe assumptions.
This work proposes and validates a novel three-stage approach to generate a patient-tailored applicator configuration without resorting to fully geometric optimisation. First, a high-potential set of dwell positions is obtained by running a modified version of the dwell time optimisation software BiCycle (Erasmus Medical Centre, Rotterdam, the Netherlands) on a grid of possible dwell positions. This results in a resolution optimal treatment plan when not considering geometry and applicator constraints. The positions with the highest dwell times according to the resolution optimal treatment plan are selected in the high-potential set. Next, a weighted set cover problem is used iteratively to find a combination of feasible needle segments to cover all dwell points in the high-potential set at a minimum distance. Lastly, needle channels to steer the needle to these segments are simultaneously optimised to be of minimum curvature and mutually collision-free.

To evaluate our approach, a virtual configuration and planning study was performed in a cohort of 22 locally advanced cervical cancer patients previously treated with the Venezia (Elekta AB, Stockholm, Sweden). The resulting treatment plans of the clinically used configuration were compared with the resulting treatment plans of the proposed patient-tailored and grid configurations on clinically relevant dose-volume histogram parameters, dwell times, conformity index and number of interstitial needles. Statistical significance is assessed with Wilcoxon signed-rank tests.

The proposed workflow was demonstrated to be feasible, and for every patient, a configuration could be generated in clinically acceptable time. All treatment plans generated for the grid configuration, the patient-tailored configuration and the clinical configuration were acceptable following the EMBRACE ll aims. Planning aims, however, were met more frequently with both the grid configurations (145/151 instances) and the patient-tailored configurations (137/151 instances) in comparison with the clinically used configurations (119/151 instances). The treatment plans generated with the grid configurations obtained significantly better (p < 0.01) median normalised CTVIR D98 dose with respect to the clinical configurations. Moreover, with the grid configurations, there was a significant improvement in median normalised D2cm3 dose for the bladder (p < 0.001), rectum (p < 0.001), sigmoid (p < 0.01) and bowel (p < 0.001) compared to treatment plans obtained with the clinical configurations. The treatment plans generated with the patient-tailored configurations obtained comparable target doses, but median normalised D2cm3 doses for the bladder (p < 0.001), rectum (p < 0.01) and bowel (p < 0.01) were significantly better for the patient-tailored configurations compared to the clinical configurations.

The proposed automated patient-tailored BT source channel configuration planning method was demonstrated to be clinically feasible. The resulting treatment plans have dosimetric advantages over the treatment plans generated with the clinical applicator configuration. Improvements to the intracavitary dwell position placement are expected to further increase dose conformity.

While learning-based control techniques often outperform classical controller designs, safety requirements limit the acceptance of such methods in many applications. Recent developments address this issue through Certified Learning (CL), which combines a learning-based controller with formal methods to provide safety guarantees. This thesis focuses on the CL based on Control Barrier Functions (CBFs), as CBFs have been widely used for safety-critical systems. However, it is non-trivial to design a CBF. Utilizing neural networks as CBFs has
shown great success, but it necessitates their certification as CBFs. In this work, we leverage bound propagation techniques and the Branch-and-Bound scheme to efficiently verify that a neural network satisfies the conditions to be a CBF over the continuous state space. To accelerate training, we further present a framework that embeds the verification scheme into the training loop to synthesize and verify a neural CBF simultaneously. In particular, we
employ the verification scheme to identify partitions of the state space that are not guaranteed to satisfy the CBF conditions and expand the training dataset by incorporating additional data from these partitions. The neural network is then optimized using the augmented dataset to meet the CBF conditions. We show that for a non-linear control-affine system, our framework can efficiently certify a neural network as a CBF and render a larger safe set than state-of-the-art neural CBF works. We further employ our learned neural CBF to derive a safe controller to illustrate the practical use of our framework.

Mobile manipulators, which combine a mobile platform with a robotic arm, are versatile robots that can be used for a variety of tasks like logistic pick-and-placing, manufacturing or assembly. Compliant control for mobile manipulators could improve the safety of the users sharing their workspace with these robots. The two general methods of compliant control, admittance control and impedance control, require force/torque sensors, which are often not available on low-cost or lightweight robots. This report presents an adaption of impedance control, including a strategy to compensate for joint friction, that can be used on current-controlled robots without the use of force/torque sensors. A calibration method is designed for the arm, that enables estimation of the actuator's current/torque ratios and frictions, used by the adapted impedance controller. Software is developed to use the controller on a combination of the Kinova GEN3 Lite arm and the Clearpath Dingo Omnidirectional base. Real-world experiments with the arm show that the calibration method is consistent and that the designed controller is compliant while also being able to track targets with five-millimeter precision when no interaction is present. Experiments with the complete mobile manipulator showcase two new modes, both suitable for interaction with a human. The first is a guidance mode where the user can control the robot using interaction with the arm. The second is a tracking mode where the mobile manipulator tracks a moving target while still being compliant.

Deep reinforcement learning presents a compelling approach for the exploration of cluttered 3D environments, offering a balance between fast computation and effective vision-based navigation. Yet, the use of 3D navigation for learning-based information gathering remains largely unexplored. Navigation in 3D space poses the challenge of having an increased state space but also provides possibilities due to the agent's increased mobility, so it is an interesting direction for research. Furthermore, current approaches to target mapping with 3D navigation do not consider cluttered environments, failing to address obstacle avoidance and occlusion handling.

This research introduces a novel deep reinforcement learning policy for vision-based information gathering with quadcopters, enabling efficient exploration of cluttered 3D environments. The core challenge is learning a time-efficient and collision-free exploration strategy, for which the policy design and the training procedure have been successfully developed. We formulate a target-searching task, where the goal is to reduce the agent's uncertainty about the target state. To achieve this goal, our method combines vision-based reasoning by deep reinforcement learning and probabilistic target mapping with an information-theoretic rewarding scheme to obtain a policy that makes informed exploration decisions.

Experiments comparing our method with a privileged greedy baseline show that in all tested environments, our policy achieves a significant outperformance. The results from our ablation study further validate our policy design, as every ablation performed results in worse exploration performance. Generally, our policy shows intelligent behaviour by effectively navigating through rooms and around obstacles. However, improvements can still be made, since failure cases where the agent gets stuck, can sporadically occur. Still, overall, the findings prove the feasibility of learning a 3D navigation policy for effective target mapping with quadcopters.

Unmanned aerial vehicles (UAVs), particularly quadcopters, are increasingly employed in diverse applications due to their manoeuvrability and affordability. However, their susceptibility to GPS jamming and the need for skilled operators limit their operational resilience. This paper presents a new design of an Image-Based Visual Servoing (IBVS) control system for autonomous quadcopter interception, utilizing only a monocular camera and an Inertial Measurement Unit (IMU). The system features a Multi-Axis PID Controller with acceleration limiting and a virtual plane projection method to decouple pitch and vertical motions, thereby enhancing interception accuracy and maintaining target visibility within the drone’s field of view (FOV). Furthermore, a perception module is designed to run Yolo-v8n and CSRT in parallel, followed by a Kalman filter, to deliver an accurate and robust representation of the target within the image, operating at 18 frames per second in Simulation-in-the-Loop (SITL) environments. Comprehensive SITL experiments and comparative analyses against recent IBVS algorithms demonstrate that the proposed proposed system achieves a 19% reduction in circular error probable (CEP) for static targets and superior performance in diverse scenarios involving moving targets. These findings validate the effectiveness of the proposed IBVS approach, offering a reliable and scalable solution for autonomous UAV interception in GPS-denied environments with potential applications in security, surveillance, and conflict zones.

Logistics and transportation can greatly benefit from the use of autonomous robots, such as self-driving vehicles. Robots can help to move goods or people without human supervision. One of the main components that enable autonomous navigation among humans is motion planning. Motion planning is responsible for computing a collision-free trajectory that moves the robot to its destination, based on the perceived information. The motion planner should be efficient, robust, and safe. This thesis contributes towards this goal by investigating the design of motion planning algorithms for autonomous navigation of mobile robots near humans.

Traditional motion planners for dynamic environments have two key limitations that this thesis aims to address. First, they assume that their model of dynamic obstacles (e.g., humans) is exactly correct, capturing it with a single deterministic prediction. In practice, the robot cannot observe human intentions and must account for its uncertainty about the human's future behavior. Second, motion planners usually compute a single trajectory around an obstacle as a result of previously taken decisions without exploring alternative options. They react slowly or even fail to find a solution when unpredicted changes make this path undesirable. This results in poor planning performance in dynamic environments.

The goal of this thesis is to develop motion planners that account for the uncertainty of human motion predictions and that are consistent and robust in their decision-making in order to deal with unpredicted changes in dynamic environments. To accomplish this goal, this thesis proposes two motion planning frameworks: scenario-based and topology-driven trajectory optimization.

The first contribution of this thesis is Scenario-based Model Predictive Contouring Control (S-MPCC), a real-time capable probabilistic planning framework that incorporates any uncertainty associated with the motion predictions of dynamic obstacles. Contrary to existing methods that only account for small variations around a single predicted trajectory (unimodal uncertainty), the proposed planner accounts for multiple possible trajectories (multi-modal uncertainty). The planner therefore safely accounts for several outcomes, for instance, to express that a pedestrian may or may not cross in front of the robot.

S-MPCC bounds the probability of collision in each time step with all obstacles through Chance-Constrained Optimization (CCO). The CCO is reformulated as an optimization without uncertainty by sampling trajectories from the predicted distribution, known as scenarios. Each scenario represents a possible position of all obstacles in one time step, and the planner avoids collisions with all scenarios. This Scenario Program (SP), through a tailored linearization, can be solved efficiently online. S-MPCC therefore plans probabilistic safe trajectories independent of the underlying distribution of the uncertainty.

S-MPCC considers the probability of collision separately for each time instance in the planned trajectory. The second contribution of this thesis, Safe Horizon Model Predictive Control (SH-MPC), builds on S-MPCC to constrain the joint probability of collision with all obstacles over the duration of the planned trajectory. Existing methods that separately constrain the probability of collision in each time step (temporal marginal) and with each obstacle (obstacle marginal) lead to overly cautious motion planning when safety constraints are enforced. SH-MPC formulates a single chance constraint to bound the overall probability of collision. This CCO is reformulated as an SP where each scenario represents a possible trajectory for all obstacles. To certify the joint probability of collision with the SP, the number of scenarios that affect the motion plan needs to be identified. SH-MPC estimates this quantity at a negligible computational cost during optimization. Consequently, SH-MPC plans trajectories in real-time under generic uncertainties that are less cautious than existing methods without compromising on safety.

The probabilistic safety of S-MPCC and SH-MPC is linked to the underlying accuracy of the prediction model of the obstacles that provide the scenarios. As a third contribution, a joint prediction and planning framework, Partitioned Scenario Replay (PSR), is proposed that replays past observations of human motion as scenarios for scenario-based planning. PSR does not fit a distribution on observed data but directly uses the data as empirical evidence of the underlying uncertainty and thereby provides a real-world safety guarantee.

A key limitation of the developed scenario-based planners and other optimization-based planners is that they locally refine an initial trajectory. This initial trajectory largely determines the quality of the final trajectory, while it does not consider other options. The fourth contribution of this thesis is Topology-driven Model Predictive Control (T-MPC) that concurrently optimizes trajectories, each attempting a different way to pass the obstacles. T-MPC is composed of a guidance planner and several parallel local planners. The guidance planner identifies guidance trajectories for several distinct maneuvers, relying on results from topology to distinguish trajectories. Each local planner is composed of an existing optimization-based planner (e.g., a scenario-based planner) and an additional set of constraints that are derived from one of the guidance trajectories. The guidance trajectories are optimized by the local planners in parallel, and the results are compared to determine which trajectory gets executed. T-MPC is faster, more consistent, and safer than several state-of-the-art planners. Contrary to similar existing work, it does not rely on an explicit lane structure and therefore enables both urban driving and mobile robotic applications.

The motion planners developed in this thesis are extensively validated in simulation and in experiments with a small-scale mobile robot and a full-scale self-driving vehicle navigating among pedestrians. The robot-agnostic implementation of the proposed planners that were developed for this thesis is available open source.@en

As robotics will play a crucial role in the future of our modern societies, the need for advancements in the field is more pronounced than ever. While robots are already present in industrial settings, they are noticeably absent from dynamic environments. Dynamic environments are characterized by other moving agents, such as humans, varying tasks and high safety requirements. With the aim to deploy robots in such environments, Trajectory Generation (TG) becomes crucial. TG approaches aim to compute sequences of control commands that bring the robot from its current configuration to a desired goal state while avoiding collisions with obstacles and itself. Thus, it is directly placed between task planning, the problem of defining what high level tasks should be executed in which order, and control, the problem of executing motor commands. A good solution to TG must be fast, to cope with changes in the environment, flexible to different goal definitions, and it should promote safety. Most advancements in the field of robotics in TG focus either on manipulators or on mobile robots. However, the combination of both systems seems inevitable for deployment in human-shared environments.

TG for mobile manipulation is usually formulated as an optimization problem of a finite time horizon. This approach is known as Model Predictive Control (MPC) and is widely used in the field of autonomous driving thanks to its feasibility and stability guarantees. In Chapter 4, we present a method to bring MPC to mobile manipulation. The method formulates the TG problem for the entire kinematic chain and relies on Free Space Decomposition (FSD) for collision avoidance. This leads to reasonable control frequencies of 10Hz independent on the amount of collision obstacles in the environment. Importantly, this approach allows for coupled motion of the mobile base and the manipulator. This is beneficial in situations where synchronization of the two subsystems is crucial, such as opening doors or moving obstacles around. Despite these simplifications on the environment representations, computational costs limit the applicability of MPC to mobile manipulation as motion is not considered truly reactive and different components, such as goal attraction and collision avoidance, are not easily separable.

A recent novel approach to receding horizon control is the formulation as a purely geometric problem. Early successes in this direction, including Cartesian Impedance Control (IC) and Artificial Potential Fields (APF), led to the formulation as sets of dynamical systems on smooth manifolds in the configuration space. The framework of Optimization Fabrics (fabrics) unifies such ideas, offers stability guarantees in static environments, and results in highly reactive behavior similar to simple low-level controllers. This framework relies on non-Riemannian geometry to shape a smooth manifold of the configuration space with individual behaviors, such as collision avoidance, joint-limit avoidance, and goal attraction. In Chapter 5, we present a generalization of fabrics to dynamic environments. We refer to the resulting framework as Dynamic Fabrics (DF). The generalization uses time-parameterized manifolds to integrate moving obstacles and time-parameterized reference trajectories. The latter is especially important for long-horizon TG that may exhibit local minima. Importantly, Chapter 5 shows that the dynamic component of DF is required when coping with moving obstacles. As repulsive forces in fabrics are proportional to the approaching speed of obstacle and robot, collision avoidance in a pseudo-static fashion is not sufficient when the robot is moving slowly. Finally, we deploy the general framework of DF to several real-world settings showing the applicability of the framework to mobile manipulation. First, we present a way to integrate non-holonomic constraints into the framework. Despite loosing formal guarantees on convergence, the method is shown to be the natural extension to wheeled mobile robots characterized by non-holonomic constraints. Second, Chapter 6 presents a symbolic implementation of fabrics to achieve higher control frequencies. Symbolic implementations are possible because the framework of fabrics is based on differential equations of second order, for which a closed-form solution exists. For changing environments, obstacles states are then only concretized at runtime. Additionally, we show symbolic hyperparameters can be tuned automatically to achieve expert-level tuning performance. Third, to overcome high requirements on the perception pipeline, Chapter 7 integrates different implicit environment representations into the framework. Using Signed Distance Field (SDF) and FSD for example is widely used in mobile robotics when formulating TG as MPC. We show that the same representations can be used in fabrics while achieving faster solver times. Finally, Chapter 8 deploys a mobile manipulator controlled by fabrics in a supermarket. Dexterous manipulation is programmed using learning-from-demonstration, with fabrics as the underlying encoding. That allows to teach rather than program complicated behaviors while maintaining properties on collision avoidance.

This thesis presents insights into aspects of motion planning, advances the framework of fabrics for TG, and compares it extensively to the more commonly used method of MPC. Through the ideas presented in this thesis, we hope to encourage the usage of geometric properties of robotic systems deployed to human-shared environments. This approach does not only provide reactive TG but also may act as a compact encoding of trajectories for learning-based methods in the future.
@en

Dynamic obstacle avoidance remains a crucial research area for autonomous systems, such as Micro Aerial Vehicles (MAVs) and service robots.
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.


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.

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.

Task and Motion Planning (TAMP) has progressed significantly in solving intricate manipulation tasks in recent years, but the robust execution of these plans remains less touched. Particularly, generalizing to diverse geometric scenarios is still challenging during execution. In this work, we propose a reactive TAMP method to deal with disturbances and geometric ambiguities by combining an active inference planner (AIP) for online action selection with a proposed multi-modal model predictive path integral controller (M3P2I) for low-level control. The AIP generates online alternative plans, each of which is translated into a cost function to be sampled for the proposed method. The proposed M3P2I then uses a parallelizable physics simulator for throwing different rollouts, leading to a coherent optimal solution by averaging the weighted samples based on their costs. Our method empowers real-time adaptation of action sequences to rectify failed plans, while also computing low-level motions to address dynamic obstacles or disturbances that could potentially invalidate the existing plan.

Theoretical findings are validated in simulation and in the real world. We show that the framework exhibits reactiveness in different scenarios, including battery charging, push-pull among obstacles, and pick-place with disturbances. We show that our framework outperforms an off-the-shelf RL method in the reactive pick-place task in terms of position error and orientation error. We also show that M3P2I is generalizable in combining different constraints, such as generating hybrid motions of push and pull for the mobile robot, and grasping objects with different grasping poses. The real-world experiments show that the system exhibits reactiveness and robustness against human disturbances in a variety of manipulation tasks.

The supporting videos can be found at https://sites.google.com/view/m3p2i-aip.

Recent progress in multiple micro aerial vehicle (MAV) systems has demonstrated autonomous navigation in static environments. Yet, there are limited works regarding the autonomous navigation of multiple MAVs in dynamic and unknown environments. The challenge arises from the complexity of the motion planning problem, which requires the MAVs to coordinate with each other while avoiding dynamic obstacles. This thesis presents a novel risk-aware decentralized multi-agent motion planning framework to address this issue. For perception, we rely on a particle-based dynamic map, which utilizes particles to represent dynamic obstacles and predict future states of dynamic obstacles. Leveraging this map representation and the shared trajectory from other agents, we evaluate the future collision risk with dynamic obstacles and other agents in a coupled manner. During the planning phase, a risk-aware kino-dynamic A* algorithm tailored to the particle-based map representation is developed, ensuring dynamically feasible paths with risks under a given safe level. Subsequently, spatio-temporal safety corridors with maximum volume are optimized by inflating from path segments, taking map particles as constraints. These corridors act as constraints for the trajectory optimization problem, which is simplified to a convex optimization problem by using Bézier splines. The proposed method is thoroughly evaluated in simulation environments featuring various quantities and shapes of dynamic obstacles. Comparative results with state-of-the-art multi-agent planners that rely on precise obstacle observations demonstrate the efficiency and safety of the proposed method. Furthermore, the effectiveness of the proposed method is validated in a more realistic simulation environment with pedestrians, using depth cameras onboard for perception.

Many autonomous navigation tasks require mobile robots to operate in dynamic environments involving interactions between agents. Developing interaction-aware motion planning algorithms that enable safe and intelligent interactions remains challenging. Dynamic game theory renders a powerful mathematical framework to model these interactions rigorously as coupled optimization problems. By solving the resultant coupled optimization problems to equilibrium solutions, the game-theoretic models explicitly account for the interdependence of agents’ decisions and achieve simultaneous prediction and planning. Coupled constraints between players, such as collision avoidance, can also be handled explicitly. However, most existing game-theoretic motion planning approaches rely on known objective models of all agents. This assumption presents a key obstacle to real-world ego-centric planning applications of these methods, where only local information is available. This thesis investigates solution approaches to relax this assumption and explicitly account for the ego agent’s uncertainty about other agents’ objectives while adaptively conducting game-theoretic motion planning.

The main contribution of this work is an online adaptive model-predictive game-play (MPGP) framework that jointly infers other players’ objectives and computes corresponding generalized Nash equilibrium (GNE) strategies. These strategies are then used as predictions for other players and control strategies for the ego agent. The adaptivity of the proposed approach is enabled by differentiating through a trajectory game solver whose gradient signal is used for maximum likelihood estimation (MLE) of opponents’ objectives. Compared with existing objective inference solutions in dynamic games, the proposed approach handles general inequality constraints in games and further supports direct integration with other differentiable modules, such as neural networks (NNs). Two simulation experiments indicate that the proposed approach performs closely to solving games with known objectives and outperforms the game-theoretic and model-predictive control (MPC) baselines. Two hardware experiments further demonstrate the real-time planning capability of the planner and its real-world applicability.

In addition to this main contribution, the second contribution of this work is a variational autoencoder (VAE) pipeline built upon the proposed differentiable game solver. This contribution aims at going beyond the point estimation in the first contribution and inferring potentially multi-modal beliefs about players’ objectives based on observations. The main idea is to employ variational inference (VI) to approximate Bayesian inference of players’ objectives. The variational autoencoder (VAE) framework is utilized for amortization to avoid per-sample optimization. Initial results on a single-player example show that after training, the proposed pipeline can: (i) generate a game objective distribution that resembles the underlying training data distribution and (ii) accurately predict a narrow, uni-modal posterior objective distribution when the observation is unambiguous based on seen data in the past and (iii) generate a multi-modal belief distribution of player’s objective to capture mostly likely modes in case of high uncertainty.

Autonomous robots hold great potential for positive impacts on society by applying them to tasks that are hazardous, repetitive, or complex and difficult for humans to perform. To achieve these tasks, autonomous robots require the ability to perceive environmental changes and create corresponding motion plans, which involve a combination of perception and motion planning techniques. Building a perception pipeline that can detect all relevant details in a dynamic environment is challenging and computationally expensive. To address this issue, the raw data output of a distance-measuring sensor can be used as a perception pipeline directly, transferring most of the computational load to the motion planner. However, most motion planning techniques cannot handle this high computational load. The motion planning technique optimization fabrics offers a promising solution, which utilizes a combination of differential equations to design robot behavior with extremely low computational complexity.

This thesis proposes a method for local motion planning that combines the low computational complexity of optimization fabrics with direct sensor integration. Our goal is to develop a method that enables autonomous robots to perceive and respond to changes in their environment quickly and efficiently. Our method, called direct sensor integrated (DSI) optimization fabrics, utilizes collision-avoidance differential equations generated for each raw data point collected from a distance measuring sensor. We adapted the regular optimization fabrics method to incorporate sensor data directly, and our analytical analysis shows that direct sensor integration does not require scaling adjustments to the regular collision-avoidance differential equations. By combining optimization fabrics with direct sensor integration, DSI optimization fabrics offers a promising solution to local motion planning. This method can enable autonomous robots to handle tasks that are challenging for humans to perform without needing a complex perception pipeline.

We conducted several experiments to assess our method, utilizing a simulated LiDAR-equipped point robot. First, we show empirically that an adjustment is required to properly scale the collision-avoidance differential equations, resulting in similar collision-avoidance behavior across sensor resolution. Second, we demonstrate that DSI optimization fabrics is feasible regarding computational complexity, achieving a frequency of 23 Hz even with the maximum sensor resolution of 2048 LiDAR rays. Third, we demonstrate the effect of varying the sensor resolution on performance in multiple goal-reaching scenarios. We measure performance by monitoring the time-to-goal, total path length, minimum clearance from obstacles, and success rates. Fourth, we compare our method to regular optimization fabrics with a simulated perception pipeline in a scenario with static obstacles and a scenario with dynamic obstacles. In both scenarios, we show comparable performance regarding time-to-goal, total path length, minimum clearance from obstacles, and success rates. Finally, we showcase a real-world application of our method.

Autonomous robots have been widely applied to search and rescue missions for information gathering about target locations. This process needs to be continuously replanned based on new observations in the environment. For dynamic targets, the robot needs to not only discover them but also keep tracking their positions. Previous works focus on either searching for static targets or tracking dynamic targets given the number of targets and their initial positions. However, the prior information including targets not moving and initial target states can be difficult to obtain in reality. There are also some efforts to solve the search and tracking task jointly by switching between the search mode and the track mode or designing hybrid heuristics. But these methods cannot account for the effect of target movement during the search process, and the trade-off between search and tracking is sensitive to the heuristics.

To overcome the limitations above, in this thesis, we propose a graph formulation of the search and tracking of an unknown number of dynamic targets. The search and tracking problem is decoupled into two parts: search for undiscovered targets and track discovered ones. The search objective is modeled by minimizing the uncertainty in the environment evolving according to a diffusion mechanism and the tracking objective is formulated as minimizing the entropy of target belief distributions. Based on that, we design a novel graph neural network architecture, trained via Reinforcement Learning, that outputs the next motion primitive for the robot to collect information in the environment. We first evaluate this framework in the pure search and the pure tracking tasks. The results show that our method outperforms a variety of baselines both when searching in small and medium-scale environments, and tracking multiple dynamic targets in medium-scale environments. Then the experiments of the search and tracking task validate that our method achieves a better trade-off under equally good search or tracking performance, and scales to a large number of targets.