Research on Multi-Robot Collaborative Pollution Source Search Methods in Unknown Environments

Aiming at the problem of searching for pollution sources in unknown environments, this paper proposes a multi-robot source-finding algorithm based on fast adaptive multi-point expectation improvement. This method first accounts for collision avoidance between robots and obstacles, and uses a separating hyperplane to construct a buffered safe Voronoi cell, which improves the overall safety performance of the multi-robot system; secondly, Gaussian process regression is used to perceive the location pollution environment, and a fast adaptive multi-point expectation improvement algorithm is introduced to dynamically optimize the sampling position of the robot in the pollution source search process, reduce data redundancy in sampling, and improve the overall performance of the overall source search. Finally, a multi-robot incremental movement strategy is adopted to reduce path oscillation, achieve a smooth search process, and improve the overall source search efficiency. The simulation experiments show that this method can quickly complete the perception of the unknown pollution environment and accurately locate the location of the pollution source.