Multi-robot Collaborative Exploration Based on Adaptive Sliding Window and Extended RRT

Exploration in dangerous and complex environments is difficult for humans. Navigation in unknown environments with robots is an important task in autonomous robotics applications, and the use of robot swarms can dramatically accelerate exploration efficiency and improve system fault tolerance. The efficiency of frontier detection during exploration determines the exploration performance of individual robots while realizing effective collaboration can reduce repeated exploration and strengthen the advantages of robot swarms. An adaptive sliding window is introduced in the sampling-based scheme, guiding frontier detection through the robot's perception of the environment. This paper further reduces the gain of the frontier around the assigned frontier points based on heuristics and utilizes the Hungarian algorithm to achieve collaboration by distributing multiple robots throughout the environment. We evaluate the proposed method in simulation and verify its efficiency relative to existing multi-robot exploration algorithms, Multi-Robot RRT-based Map Exploration Algorithm (MRRBME). The results show that explore the same environment, the proposed method has up to 17.9% shorter path lengths and up to 19.0% shorter exploration times compared to MRRBME.