面向观测融合和吸引因子的多机器人主动SLAM

PDF下载HTML

Title:

Multi-robot active SLAM for observation fusion and attractor

作者:

王贺彬¹, 葛泉波², 刘华平³, 袁小虎⁴

1. 杭州电子科技大学 自动化学院，浙江 杭州 310018;
2. 同济大学 电子与信息工程学院，上海 201804;
3. 清华大学 计算机科学与技术系，北京 100084;
4. 清华大学 自动化系，北京 100084

Author(s):

WANG Hebin¹, GE Quanbo², LIU Huaping³, YUAN Xiaohu⁴

1. School of Automation, Hangzhou Dianzi University, Hangzhou 310018, China;
2. School of Electronics and Information Engineering, Tongji University, Shanghai 201804, China;
3. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China;
4. Department of Automation, Tsinghua University, Beijing 100084, China

关键词:

主动同时定位与建图; 多机器人协作; 吸引因子; 凸组合融合; 扩展卡尔曼滤波器; 最优控制; 互信息; 多目标优化

Keywords:

active simultaneous location and mapping; multi-robot cooperation; attractor; convex combination fusion; extended Kalman filter; optimal control; mutual information; multi-objective optimization

分类号:

TP510.80

DOI:

10.11992/tis.202006019

摘要:

针对未知环境下多机器人主动SLAM(simultaneous localization and mapping)存在不能完全遍历环境、定位精度不理想等问题，本文基于EKF-SLAM(extended Kalman filter-simultaneous localization and mapping)算法提出一种多机器人主动SLAM算法。通过引入吸引因子，增强多机器人系统之间的交流，提升机器人自身定位精度与环境建图精度，同时又引导多机器人团队进行探索环境。当同一地标被多个机器人观测到，采用凸组合融合方法融合各个机器人对地标的估计，从而降低被估计地标的不确定度。仿真结果表明，所提算法能够对环境进行覆盖遍历，提升对地标估计的定位精度。

Abstract:

Because multi-robot active SLAM cannot fully traverse an environment, and the localization accuracy is not ideal in an unknown environment, a new multi-robot active SLAM algorithm is proposed in this paper. By introducing attractors to enhance communication between multi-robot systems, the accuracy of robot localization and mapping is enhanced, and multi-robot teams are guided to explore the environment. When the same landmark is observed by multiple robots, convex combination fusion is used to fuse the estimate of the landmark by each robot, thereby reducing the uncertainty of the landmark. The simulation results show that the proposed algorithm can cover and traverse the environment and improve the localization accuracy of landmark estimation.