基于进化梯度搜索的机器人主动嗅觉仿真研究

提出了基于进化梯度搜索的多机器人主动嗅觉的一种实现策略．首先用Fluent软件建立了一个时变的气态流体环境；其次给出了在此仿真环境中的基于进化梯度搜索的机器人主动嗅觉实现过程，包括发现气体、跟踪气体和气味源确认．为了弥补进化梯度搜索法在机器人数量有限情况下存在的不足，本文算法还使用了风向信息．仿真结果验证了该搜索策略的有效性．通过与传统的基于单机器人的浓度梯度搜索策略比较，验证了本文所用方法的优越性．

Simulation Study on Robot Active Olfaction Based on Evolutionary Gradient Search

An active olfaction implementation scheme based on evolutionary gradient search (EGS) using a swarm of robots is put forward.Firstly,a time-variant airflow environment using Fluent software is set up.Secondly,the EGS-based robot active olfaction implementation procedure in the simulated environment is presented,including plume finding,plume tracking and odor source declaration.To overcome the drawback of the EGS algorithm,i.e.its performance will degrade when the number of robots is not big enough,the information of wind direction is also used.Simulation results validate the proposed search scheme.Comparing with the traditional concentration-gradient based search methods using single robot,advantages of the EGS-based method are demonstrated.