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.

With the world’s population recently surpassing the 8 billion mark, population growth poses significant challenges on the planet. This growth is particularly evident in urban areas, and as a consequence, cities must find innovative ways to accommodate the increasing pressure on the current road infrastructure. In Amsterdam, a city with an extensive network of urban canals, Autonomous Surface Vessels (ASV) could play an important role in alleviating the pressure on the road by transporting goods and people through the canals. However, autonomous navigation in narrow and unstructured canals amongst human-operated vessels is still a great challenge. Recently, a sampling-based Interaction-Aware Model Predictive Path Integral (IA-MPPI) control framework was proposed for this purpose. The method pro- vides trajectories for all agents in a local multi-agent system based on their current state and estimated local goals. This thesis builds upon this framework by integrating a learning-based trajectory prediction model to provide better estimates of the local goals of interacting agents. We adapt a state-of-the-art trajectory prediction model and train it on simulated vessel data to predict vessel trajectories. We then provide heuristics to extract local goals from these predictions and integrate them into the IA-MPPI framework. With extensive experiments in simulated environments of Amsterdam’s canals, we show that our proposed method outperforms the baseline in high-interaction scenarios and achieves similar performance to the method with communication capabilities. Furthermore, we provide insights into the benefits of interaction-aware planning over planning with fixed trajectory predictions for other agents’ motion. In additional experiments, we provide heuristics to use the trajectory predictions to improve the sampling distribution of the IA-MPPI. Preliminary results of this method provide insights into the potential benefits of this approach and we suggest directions for future research in this area.