多机器人协同的SLAM算法研究

SLAM algorithm for multi-robot collaboration

针对传统的室内多机器人SLAM算法存在探索任务分配灵活低,重叠度高,导致地图融合和建图精度不高的问题。设计一个基于PSO算法的多机器人协同建图与路径规划系统。首先,采用SLAM系统中的Gmapping算法作为基础算法,加入PSO算法将地图融合问题转化为最优求解问题,即找到两个地图重叠度最高的转换矩阵,实现地图融合和多机器人协同建图。结果表明,相同室内环境下,单机器人的平均探索时间为216 s,探索覆盖率为73..38%;而双机器人的平均探索时间仅为47 s,比单机器人的探索时间低了169 s;且双机器人的探索覆盖率为99.69%,比单机器人高出了26.31%。由此说明,双机器人的探索效率和探索覆盖率更高。对比于现有的EKF-CSLAM算法和基于因子图地图融合算法,本算法的建图精度高达99.75%。地图融合和建图精度明显更佳,进一步说明提出的融合算法可提升多机器人室内环境协同建图的效率和鲁棒性。

For the traditional indoor multi-robot SLAM algorithm has low flexible task assignment and high overlap,which leads to low map fusion and low mapping accuracy.A multi-robot collaborative mapping and path planning system based on PSO algorithm is designed.First,the Gmapping algorithm in SLAM system is adopted as the basic algorithm,and the PSO algorithm is added to transform the map fusion problem into the optimal solution problem,that is,the conversion matrix with the highest overlap between the two maps is found to realize map fusion and multi-robot cooperative map construction.The results show that in the same indoor environment,the average exploration time of single robot is 216 s and the exploration coverage is 73.38%;the average exploration time of double robot is only 47 s,169 s lower than that of single robot;and the exploration coverage of double robot is 99.69%,26.31%higher than that of single robot.This shows that the exploration efficiency and exploration coverage of the dual robots are higher.Compared with the existing EKF-CSLAM algorithm and factor-based map fusion algorithm,the accuracy of this algorithm is 99.75%.Map fusion and mapping accuracy is significantly better.It further shows that the proposed fusion algorithm can improve the efficiency and robustness of multi-robot indoor environment collaboration.