Abstract:In this paper, we proposed a novel multi-robot collaborative exploration method to improve the efficiency and robustness of multi-robot exploration in unknown environments. Firstly, a novel frontier detection algorithm based on hybrid multi-strategy rapidly-exploring random tree (HMS-RRT) is proposed, which is composed of an adaptive incremental distance strategy, a subregion sampling strategy and a greedy frontier-based exploration strategy. To improve the frontier detection performance of the algorithm, we adopt the Voronoi diagram to continuously partition the explored region, and dynamically adjust the incremental distance according to the density of obstacles in the subregions. To avoid the algorithm is trapped in the local optimum, we use Gaussian distribution to calculate the sampling probability in each subregion, so that the algorithm tends to sample in the subregion with lower crowded level of nodes and cover the unexplored regions quickly. Secondly, we introduce the greedy frontier-based exploration strategy to explore all Voronoi polygons in turn and refine the search results, meanwhile, the centroid of each frontier region is extracted as the exploration target point. Then, a multi-robot task assignment strategy based on improved market mechanism is introduced to dynamically assign the exploration target points to each robot, and the map-merging algorithm is used in the exploration process to merge several local maps in real-time. Finally, an experimental testing platform is developed based on Robot Operating System (ROS) and a series of experiments are carried out. The results show that our method can improve the efficiency and reliability of multi-robot exploration in both the simulations and the prototype experiments.

Bio-inspired Decentralized Multi-robot Exploration in Unknown Environments

Meeting-Merging-Mission: A Multi-robot Coordinate Framework for Large-Scale Communication-Limited Exploration.

Moving Forward in Formation: A Decentralized Hierarchical Learning Approach to Multi-Agent Moving Together

Mapless Collaborative Navigation for a Multi-Robot System Based on the Deep Reinforcement Learning

Optimal Exploration Algorithm of Multi-Agent Reinforcement Learning Methods (Student Abstract)

Cooperative Search and Exploration in Robotic Networks

Distributed multi-robot potential-field-based exploration with submap-based mapping and noise-augmented strategy

PIPO: Policy Optimization with Permutation-Invariant Constraint for Distributed Multi-Robot Navigation

A Pso-Inspired Multi-Robot Map Exploration Algorithm Using Frontier-Based Strategy

Decentralized Strategy for Cooperative Multi-Robot Exploration and Mapping

Multirobot Unknown Environment Exploration and Obstacle Avoidance Based on a Voronoi Diagram and Reinforcement Learning

SMMR-Explore: SubMap-based Multi-Robot Exploration System with Multi-robot Multi-target Potential Field Exploration Method.

Learning to Cooperate in Decentralized Multi-robot Exploration of Dynamic Environments

Bayesian Reinforcement Learning for Multi-Robot Decentralized Patrolling in Uncertain Environments

Frontier-based multi-robot map exploration using Particle Swarm Optimization.

A Multi-agent Reinforcement Learning Method for Swarm Robots in Space Collaborative Exploration

Learning to Cooperate Via an Attention-Based Communication Neural Network in Decentralized Multi-Robot Exploration.

HMS-RRT: A novel hybrid multi-strategy rapidly-exploring random tree algorithm for multi-robot collaborative exploration in unknown environments

Multi-robot Cooperative Navigation Method based on Multi-agent Reinforcement Learning in Sparse Reward Tasks

A new method of multi-robot cooperative exploration