This thesis addresses the detection and prevention of adversarial attacks on self-triggered control (STC) systems and demonstrates how manipulating the sampling times can be used to provide a secure control framework. Secure control is vital to guarantee both the safety and security of critical infrastructure, which has been the target of malicious attacks over the past decade.

First, a novel watermarking scheme is proposed based on the early triggering of a well-designed STC policy, such that stability is guaranteed to be preserved. We show that this watermarking scheme, together with an event-triggered χ² detector we design, is able to detect replay attacks. An online heuristic for obtaining an optimal early triggering policy is provided. If certain assumptions hold we show that the early triggering policy is a discrete uniform one. Through an illustrative example, both a quantitative and qualitative comparison between two other watermarking schemes are provided. We conclude that none of the watermarking schemes can claim absolute superiority, and trade-offs between all considered schemes exist.

Next, we propose a new type of attack called a switched zero dynamic attack (ZDA), and provide an algorithm on how to construct these switched ZDA. We show that certain STC systems are susceptible to such attacks, and demonstrate that by tuning the triggering parameters there exist sufficient conditions such that these attacks are no longer disruptive. The effect of additive perturbations and a non-zero initial condition, as well as the proposed tuning method, are shown in a numerical example. We provide a qualitative comparison between several other countermeasures in the literature, which we extend for aperiodic sampling when needed. Finally, shortcomings and future directions are discussed.

This work builds upon the Gaussian Belief Propagation (GBP) stack, utilizing it as the core framework for distributed multi-robot exploration missions. The GBP stack’s key strength lies in its single-factor graph representation of competencies such as planning, map consensus, and task allocation, making it a powerful tool for intelligent and collaborative robotic behavior. However, previous applications of GBP stack were limited to open environments without obstacles. To address this limitation, we extend the GBP stack by representing the environment as a navigational graph, allowing for task allocation and global path planning in more complex, obstacle-filled environments. This enhancement integrates seamlessly with the single-factor graph approach, enabling both navigational tasks and symbolic tasks, such as opening doors or performing inspections, to be efficiently distributed among a heterogeneous fleet of robots. A feature absent in earlier versions. Simulation results demonstrate improved task coordination, exploration efficiency, and adaptability to varied tasks, showcasing the potential of this extended framework for more challenging multirobot exploration scenarios.

Simulation environments are useful for a wide range of applications and their functionalities continue to improve every year. The aim of this thesis project was to create a simulation environment with high levels of realism and assess its capabilities through the use case of generating distributed drone traffic rules.

This thesis project was executed at the company Almende in cooperation with its sub-company DoBots. The distributed drone traffic rule generation use case was a contribution to the ADACORSA project which Almende takes part in. The goals of the thesis were the following. A simulation environment incorporating elements of realism was created using the software tools Almende has experience with. The capabilities of this environment were tested through the said use case. According to company requirements, this process was realised using an incremental complexity approach, and it utilized machine learning techniques. It was expected that the incremental complexity approach would help speed up the learning process of more complex scenarios, while machine learning would provide the benefit of smart automation and impartiality to solutions. For this process, a model was also devised for creating such rules, incorporating the incremental complexity approach and a standard machine learning technique. Finally, the benefits of the incremental complexity approach were also assessed.

The four building blocks of the implemented simulation environment were Gazebo, PX4, ROS2 and the iris drone model. The code was built upon the OpenAI ROS software package, with extensive modifications. The main work done during the implementation consisted of multiple parts. The original software package was cleaned of unnecessary elements, the PX4 software and the iris drone model were integrated into the environment and the simulation speed was increased. The whole setup was migrated from ROS to ROS2. The use of multiple drones in the same environment was also solved. An implementation of the deep Q-learning algorithm, which was selected as the learning method, was added. Different scenarios were created, as well as some logging and visualization tools. Additionally, the system was set up to run in docker containers and was highly parameterized for future use.

Different elementary traffic scenarios were implemented in the simulation and then executed. Based on these the following points were determined. It was concluded that the created simulation environment was well-fitted for running scenarios for traffic rule generation. It was also suitable for executing the scenarios using the incremental complexity approach, although its usability also depended on how the neural networks of the scenarios were set up. The environment was also feasible to work with machine learning from an engineering viewpoint, however, it did not perform well for its scientific research. This was due to not having statistically well-supported results, which was caused by the long time it took to execute a scenario. The benefits of the incremental complexity approach were also examined. For the simple case tackled in this project, it was found that the incremental complexity approach provided no significant advantages with regard to learning speed.

In robotic manipulation of deformable objects, the continuum nature of the object state leads to prohibitively high degrees of freedom when traditional modelling techniques are applied, leading to much research focusing on approaches that avoid an explicit object model. However, without such a model the opportunities for long horizon planning over the object shape or the application of optimal control methods are limited. In this work, we take a step to address this gap for the category of deformable linear objects by developing an accurate, low-dimensional manipulator-object model using the Affine Curvature strain parametrisation. We derive the model kinematics and dynamics, and detail a procedure for experimentally determining the unknown material parameters, using heavy electrical cables as a realistic test case. We then prove the viability of the model by experimentally demonstrating accurate feedforward control of the object in steady state, allowing for a significantly expanded reachable workspace and control of the endpoint orientation.

Socially compliant robot navigation in pedestrian environments remains challenging owing to uncertainty in human behavior and varying pedestrian preferences in different social contexts. Local optimization planners like Model Predictive Control can incorporate collision avoidance constraints, but they can only lead to socially compliant trajectories if the cost function embeds information about the desired social behavior. The same holds for Reinforcement Learning, where a sophisticated reward function needs to be defined. However, formalizing social behavior through a reward or cost function is difficult due to the complex nature of pedestrian behavior. Imitation learning allows for inferring the desired behavior by learning from human demonstrations, making them suitable for learning socially compliant navigation policies but without any safety considerations. In this work, we propose to learn a socially compliant navigation policy directly by observing surrounding pedestrians’ trajectories from a commonly available detection and tracking pipeline and combine it with a local optimization planner to enhance safety. A Subgoal Recommender policy is developed to guide the local optimization planner to generate socially compliant trajectories by providing intermediate subgoals. To adapt the policy to changing social contexts without forgetting previously learned information, we train the Subgoal Recommender in a Continual Learning setup exploiting new pedestrian data.
We demonstrate in simulation that our method can learn a policy that has similar performance metrics as that of the observed trajectories with 95% confidence estimated from a t-test, resulting in a lesser number of collisions. Further, the policy can adapt to different social preferences exhibited by pedestrians, while being able to remember the learned behavior in a previously encountered social context. Furthermore, we show that our proposed method can learn navigation policies from actual pedestrian data recorded using the onboard perception pipeline of a Clearpath Jackal robot.

There is growing interest to control cyber-physical systems under complex specifications while retaining formal performance guarantees. In this thesis we present a framework for formal control of uncertain systems under complex specifications. We consider dynamical systems with random disturbances, whose probability distribution is unknown. When it comes to the specifications, we focus on those given as syntactically co-safe linear temporal logic (scLTL) formulas. Such formulas resemble natural language, and allow us to reason over complex behaviours of the system.

We follow an abstraction-based approach: we abstract the original system to a finite-state Markov model, in which the state discretization error as well as the distributional ambiguity, are embedded as uncertainties. To do so we make use of tools from optimal transport and ambiguity sets of probability distributions. After that, we synthesize a strategy for the abstraction and obtain probabilistic guarantees that the abstraction satisfies the specifications. Finally, we correctly refine the strategy to one that the original system can follow, and prove that the guarantees we obtained for the abstraction also hold for the original system.

We propose two approaches to obtain the abstraction. First, we propose a data-driven approach to abstract the system into an interval Markov decision process (IMDP) when samples from the unknown distribution is available. Then we use already existing algorithms to synthesize a strategy for the IMDP. Secondly, we propose an approach to abstract the original system into a robust Markov decision process (robust MDP). This second approach is applicable to more general uncertainty models besides the data-driven one, and reduces conservatism of the abstraction. Furthermore, we propose an algorithm to synthesize robust strategies for robust MDPs, which also renders the guarantees that the abstraction satisfies the specifications. Finally, we demonstrate the usefulness of our proposed approaches through several case studies that involve both linear and nonlinear systems.

Motion planning in cluttered environments is challenging for multi-robot systems, in which each robot needs to avoid obstacles as well as other robots. This thesis presents a distributed risk-aware motion planning method for multi-robot systems in dynamic environments. For each robot navigating in a multi-robot scenario, two major risk elements are considered and formalized: a) the collision risk that is assessed using the defined "deformed distance to the centroid of free space" metric, and b) the congestion risk that is assessed via the designed "potential to goal" metric. These risk elements are incorporated into a distributed model predictive control (MPC) framework for risk-aware multi-robot motion planning, in which the collision and congestion risks of each robot are minimized. Simulation results show that the proposed method can improve the robot's safety regarding clearance to each other and obstacles comparing to the baseline method without risk minimization. Moreover, the trajectory efficiency, i.e., time to reaching goals, is also improved thanks to minimizing the congestion risk. We also validate the proposed method in real experiments with a team of Crazyflie 2.1 quadrotors.

The concept of motion planning for a path following task is highly relevant. Having a comfortable reference motion is an advantage for automated vehicles, since this leads to more higher comfort for the passengers, in turn leading to higher adoption of ADAS. Defining a speed profile can be done in three different ways. First is a heuristics based methods, such as seen in IPG CarMaker's driver model. Secondly, line-of-sight based methods use the knowledge that blind corners can contain potential moving obstacles, leading to slower speeds to compensate for the braking distance. Finally, optimization based methods, such as laptime optimization, can be used by defining dynamics, constraints and a cost function to find the optimal motion for the given scenario. Comfort is primarily a function of acceleration and jerk. The RMS of the acceleration should be as low as possible for more comfort, while jerk lower than a determined value can be ignored from a comfort point of view. Finally, the scope and the contributions to the state of the art have been brought forward. A curvilinear approach in combination with a distance domain approach is utilized, since this allows for only the curvature of the road to be used as a reference. To generate a speed profile longitudinal point mass dynamics are utilized with the jerk as an input to the optimization. The maximum speed is determined beforehand using the curvature of the road and assuming steady-state cornering. For each point along the road a speed is now determined that should not be exceeded. Generating a speed profile using a linear optimization can not be done accurately. However, an assumption can be made that the vehicle takes the same time for each distance step, leading to an approximate solution. The non-linear optimization is able to overcome the shortcomings of the linear optimization, by including the time as a state in the solution. By incorporating the lateral dynamics in the EOM the predetermined maximum speed is no longer needed. The lateral dynamics are simplified by assuming the vehicle is always aligned with the reference, therefore ignoring yaw dynamics. Leading to a complete description of the longitudinal and lateral motion that can be solved using a non-linear optimization method. ACADO is used for the non-linear optimization. A benchmark is performed against a state of the art literature example based on a non-linear optimization method that is able to minimize motion sickness based on vibrations experienced by the passengers. The speed profile generation based on linear optimization is not able to reproduce the resulting speed profile as seen in the example. For the non-linear optimization method the speed profile obtained for the time optimal cost function is very close, however for the comfortable motion, the speed through the corners was still too high to be considered comfortable. By minimizing the acceleration and jerk the motion planning system including the lateral motion is able to accurately reproduce similar speed profiles for the three different cost functions of the example. In Simcenter Prescan a complex 10 degrees of freedom vehicle model containing both sprung and unsprung mass has been presented. A driver model, based on two PD controllers for longitudinal control and the Prescan built-in path follower for lateral control, has been introduced. The motion planning containing both lateral and longitudinal motion has been used to define a comfortable reference speed and trajectory. The simulation results show that the reference motion is comparable to the resulting motion, both in the longitudinal and the lateral direction. Indicating that the motion planning system produces accurate results. On the other hand it is also observed that the longitudinal control of the vehicle needs a more complex control scheme. For the lateral control a predictive control scheme is needed to be able to follow the road with low lateral offset.

In the future, Human Driven Vehicles (HDV's) are expected to interact with Automated Vehicles (AV’s) and Connected AV’s (CAV’s). Due to the differences in the expected driving behavior of AV’s (ACC) and CAV’s (CACC) compared to HDV’s, the nature of traffic breakdown phenomena in the future can be expected to change. AV’s (ACC) and CAV’s (CACC) refer to AV’s enabled with Adaptive Cruise Control functionality and CAV’s enabled with Co-operative ACC functionality respectively. Currently, there are traffic management measures which address traffic breakdown for the current situation. With the expected changes in breakdown phenomena in the future, will the current measures be effective in addressing the different nature of breakdown in the future? This research answered this question through simulation (Vissim), by focusing into the effectiveness of one of the current measures. The current measure whose effectiveness was analyzed is Variable Speed limits (VSL) applied through the concept of feedback Mainstream Traffic Flow Control, MTFC-VSL, at on-ramp merge sections. Before conducting simulations, it was hypothesized that MTFC-VSL control effectiveness in addressing traffic breakdown increases as CAV’s (with ACC and CACC functionality) penetration rate increases in mixed traffic, because CAV’s can be expected to precisely follow the speed limits. Mixed traffic in this research comprised of HDV’s and CAV’s (with ACC and CACC functionality). CAV’s (with ACC and CACC functionality) implies that CAV’s majorly differ with that of HDV’s in car following behavior and not in lane change behavior. Simulation results and analysis revealed that the hypothesis doesn’t hold good. Improvements in average Travel Time (TT) of mainline vehicles and average network speed due to the presence of MTFC-VSL control compared to the absence of it, deteriorated as penetration rate of CAV’s (with ACC and CACC functionality) increased until 20% in mixed traffic. For further penetration rates the improvements fluctuates. On-ramp vehicles for most of the scenarios of mixed traffic, are better off without MTFC-VSL control as the presence of it increases the vehicles average TT. MTFC-VSL control doesn’t effectively address the capacity drop phenomenon for various scenarios of mixed traffic. Lastly, it was found that for less than 20% CAV’s (with ACC and CACC functionality) penetration rate in mixed traffic, MTFC-VSL control effectiveness can be expected to overall increase if Intelligent Speed Adaptation is installed as an On-Board Unit in HDV’s, as it limits HDV’s exceeding the speed limits.It must be noted that, MTFC-VSL control was set up considering the practical considerations of implementing in real life, which can also be expected to play a significant role in hypothesis not being valid. Given the ineffectiveness of MTFC-VSL control for various scenarios of mixed traffic, the future traffic management measures should focus on the causes of breakdown phenomena which aren’t addressed by MTFC-VSL control. One of the proposed measures is a combination of a merging assistant strategy & MTFC-VSL control to better address traffic breakdown than MTFC-VSL alone. Merging assistant strategy utilizes the connectivity feature of CAV’s to foster smoother merging of on-ramp vehicles which isn’t addressed by MTFC-VSL control.

The general approach to generate collision free motion in a constraint environment is to use path planners, which demand a known environment and potentially fail otherwise. Learning from Demonstration (LfD) can be used instead to teach the robot unknown parts of the environment, such as a goal deviation or an unforeseen obstacle. The general approach is to train a model offline and expect it to perform well afterwards. Problems arise however when the model is trained insufficiently or unknown variations have occurred in the environment, which demand for refinement of the model. In online learning the operator is allowed to refine a predicted trajectory during execution time, where the state-of-the-art methods focus on kinaesthetic teaching. The contribution of this research is the development of a teleoperated online learningmmethod, where the operator can make refinements by moving a haptic stylus (Phantom Omni) in the desired direction. By doing this, a force is felt proportional to the magnitude of refinement. After creating such a refined trajectory, it is used to update a condition dependent probabilistic trajectory model. The proof of concept was shown on a 2D example and on a simulated robot that shows that we can adapt an initial model when unknown variations occur and that the method is able to deal with different object positions and initial end effector poses. To show if other people can use the method, a human factors experiment is performed, comparing the developed method against three other methods on how much time it takes to successfully adapt a model (refinement time) and on the perceived workload. Two different parameters are varied, which are the teaching device (stylus or keyboard) and the learning mechanism (online or offline). The expectation was that both online with stylus has the lowest refinement time and workload, but the results show that only online has a significant improvement over offline methods (p = 7.94 × 10^−12 and p = 0.000512 respectively). This is explained by the fact that only small corrections have to be made and in a maximum of three degrees of freedom (DoFs). No significant difference was found between keyboard and stylus (p = 0.755 and p = 0.302 respectively). An explanation for this is that this is task, person and implementation dependent. The recommendations are to evaluate the proof of concept on the real robot and to extend the method with orientation refinement, such that more complex tasks can also be dealtwith. We hypothesize that with these tasks the combination of online with stylus does perform the best.

The free energy principle is a recent theory that originates from the neuroscience. It provides a unified framework that combines action perception and learning in the human brain. This research aims to implement the perception aspect of the free energy principle into robotics. This is achieved via the dynamic expectation maximisation (DEM) algorithm. DEM is derived from the free energy principle and provides a novel solution for state and input observation for LTI systems under the influence of coloured noise. This thesis provides an overview of how the DEM observer is derived from the free energy principle. Thereafter, an experimental design is presented in which data is collected to validate the performance of DEM on experimental data. The data is collected from a quadcopter drone flying in wind conditions. A detailed overview of the DEM observer settings is presented and motivated. \\ This research shows that because of its use of generalized coordinates DEM is able to leverage the coloured nature of the noise for better state estimation. This is demonstrated by the fact that DEM obtains a higher state estimation accuracy than other coloured noise state observers, such as state augmentation and SMIKF. Moreover, in input estimation, DEM is able to obtain similar results as an unknown input observer. Finally, the accuracy vs complexity trade off of DEM is highlighted.

Nonholonomic mechanical systems encompass a large class of practically interesting robotic structures, such as wheeled mobile robots, space manipulators, and multi-fingered robot hands. However, few results exist on the cooperative control of such systems in a generic, distributed approach. In this work we extend a recently developed distributed \ac{IDA-PBC} method to such systems. More specifically, relying on port-Hamiltonian system modelling for networks of mechanical systems, we propose a full-state stabilization control law for a class of nonholonomic systems within the framework of distributed \ac{IDA-PBC}. This enables the cooperative control of heterogeneous, underactuated and nonholonomic systems with a unified control law. This control law primarily relies on the notion of Passive Configuration Decomposition and a novel, non-smooth control law proposed here. A low-level collision avoidance protocol based on the \ac{APF} method is also implemented in order to achieve dynamic inter-agent collision avoidance, enhancing the practical relevance of this work. Theoretical results are tested in different simulation scenarios in order to highlight the applicability of the derived method.