改进粒子群优化算法的多机器人地图构建

针对大规模的未知环境,对一种SA-PSO(Simulated Annealing-Particle Swarm Optimization)算法的多机器人构建地图方法进行研究。多机器人构建地图,即将多个机器人建立的局部地图融合成全局地图,可以更加高效地完成环境地图的绘制。利用粒子群优化(PSO)算法搜索局部地图之间的最优转换矩阵来进行地图配准;再根据局部地图重叠区域匹配的成功率设计自适应概率函数,即重新进行地图配准的概率;最后将配准后的局部地图融合成全局地图。该方法有效解决了PSO算法易陷入局部最优引起的地图融合失败问题,提高了地图融合的成功率。

MULTI-ROBOT MAPPING BASED ON IMPROVED PARTICLE SWARM OPTIMIZATION

Under large-scale unknown environment,we proposed a modified SA-PSO(Simulated Annealing-Particle Swarm Optimization) for multi-robot mapping.Multi-robot mapping,fusing local maps established by each robot into global map,could map environment more efficiently.PSO was used to register maps by searching the optimal transformation matrix among local maps,and then adaptive probability function,the probability of map re-registration,was designed based on the success rate of overlapping region matching in local maps.Finally,the local map was fused into a global map.This method effectively solves the problem of map fusion failure caused by the PSO falling into local optimum,and improves the success rate of map fusion.