Improved Task Assignment and Path Planning for Multi-Agent Coverage Control

Coverage control is an important issue in multiagent systems, aiming to achieve comprehensive coverage of environmental areas through the collaboration of multiple agents. Due to the dynamic changes and uncertainties in actual environments, how to efficiently perform task assignment and path planning becomes key to enhancing system performance. This work proposes an improved algorithm for task assignment and coverage path planning that addresses the coverage control problems of multi-agent systems across a spectrum of environments, ranging from the straightforward to the intricate. In less complex settings, our approach integrates a mutual exclusion-based task allocation strategy with a well-established path planning algorithm to streamline task execution and trajectory determination. Conversely, in more intricate scenarios, we propose a novel probability-improved deep Q- learning algorithm to facilitate coverage path planning amidst dynamic conditions. Through a series of simulation experiments, this paper demonstrates the effectiveness and superiority of the proposed methods in different environments, especially their adaptability in the face of dynamic changes and unknown environments. This research provides an efficient and feasible solution for the application of multi-agent systems in complex environments.