Multirobot unknown environment exploration and obstacle avoidance based on a Voronoi diagram and reinforcement learning

Abstract

Autonomous exploration in dynamic and unknown environments poses severe challenges for multirobot systems, requiring the consideration of key factors such as task allocation, robot coordination, and dynamic obstacle avoidance. To address these challenges, this paper proposes a high-performance collaborative exploration strategy for multiple robots that combines the Voronoi diagram with a multiobjective cost function model. This strategy coordinates robot movement through dynamic Voronoi partitioning while integrating obstacle avoidance, exploration, and coordination into a multiobjective optimization goal, ensuring efficient exploration and significantly enhancing coverage of unknown environments. Furthermore, by integrating an obstacle avoidance algorithm based on deep reinforcement learning, the proposed method successfully resolves the interference of unpredictable obstacles in dynamic environments, ensuring the system’s safety and reliability. Additionally, the introduction of transfer learning enables robots to adapt quickly to various complex environments, significantly improving learning efficiency. In multiscenario simulation experiments, the proposed methods all achieved optimal coverage and minimal collisions. Finally, tests in real-world environments further verified the effectiveness of the proposed method.