复杂环境移动群机器人最优路径规划方法  

Dynamic differential evolution algorithm for swarm robots search path planning

作　　者：徐雪松[1,2] 杨胜杰[1] 陈荣元[1] 

机构地区：[1]湖南商学院移动商务智能湖南省重点实验室,长沙410205 [2]湖南省移动电子商务协同创新中心,长沙410205

出　　处：《电子测量与仪器学报》2016年第2期274-282,共9页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金重大国际合作项目(71210003);国家自然科学基金重点项目(71431006);国家社科基金项目(14BJY066);教育部人文社科基金青年项目(12YJCZH233);湖南省教育厅科学研究青年项目(13B060);国家留学基金资助

摘　　要：研究了一类复杂环境下移动群机器人的建模与控制策略。采用栅格法对机器人工作环境进行建模,基于个体的有限感知能力和局部的交互机制设计了响应概率函数,解决群机器人任务分配与信息共享难题。通过施加螺旋控制于早期信号搜索,并将该搜索信息作为启发因子改进动态差分进化算法,对群机器人进行路径优化。仿真结果表明,当响应概率函数中距离变量调节因子β=0.006时,任务分配控制算法达到最好效果。同时,移动群机器人路径规划的平均路径长度珔S,平均移动时间珔T以及平均收敛代数珚M,相比扩展PSO算法分别提高了16%、57%及230%。最后,将该算法应用于ASUIII型轮式移动群机器人物理实验,并设计了协同控制平台,具有较好的工程应用价值。A novel optimization algorithm based on differential evolution is proposed in this paper. The modeling and the control strategies of swarming robots for search planning in a complex environment are discussed. Grid method is used for robot working environment modeling. The response probability function is designed based on individual＇s limited cognitive ability and local interaction mechanism,which can solve the problem of the swarm robot task allocation and information sharing. Robots moving spirally to search cues can offer evidence for using dynamic differential evolution algorithm to search target optimally. The simulation results show that when the response probability function distance variable regulating factor β = 0. 006,task allocation control algorithm can achieve the best effect. At the same time,the mobile robot path planning group of average path length,average moving time and average convergence algebraic extension compared to PSO algorithm is enhanced by 16%,57% and 230% respectively. This algorithm is introduced to AS-UIII wheel mobile robots real experiments and illustrated its engineering application value.

关 键 词：群机器人 动态差分进化算法 路径规划 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置] TN209[自动化与计算机技术—控制科学与工程]