POMDP模型在多机器人环境探测中的应用研究

Research on Multi-robot Environment Exploration using POMDP

为了提高多机器人环境探测的效率和精度,本文提供了一种基于部分可观马尔可夫决策过程(Partially observable markov decision process,POMDP)的路径规划方法来控制多个装有传感器的机器人实现对环境的协同探测。建立了多机器人环境探测系统的POMDP模型,以信息熵作为回报函数,令机器人沿着信息熵最大的方向不断移动。机器人对环境的信念采用非参数的、基于样本的表示,并用贝叶斯滤波来更新机器人对环境的信念。在仿真试验中,对两种环境的CO浓度进行了探测,都得到了精确的测量结果。与传统的全覆盖路径规划的方法相比,该方法在效率和精度上都具有优势。

To improve the efficiency and accuracy for exploring a multi-robot environment,this paper proposes a path planning method based on the partially observable Markov decision process(POMDP)to control multiple robots equipped with sensors and to realize the coordinated exploration of the environment.Taken information entropy as the return function,the multi-robot environment exploration system based on the POMDP is established to move the robots with the largest information gain in the direction.The robot′s belief in the environment uses a non-parametric,sample-based representation,and the Bayesian filtering is used to update the robot′s belief in the environment.With our simulation software,the CO concentration of the two environments was precepted.The exploration results are in a good agreement with the predesigned environment.Compared with the traditional full coverage path planning method,the system proposed in this paper has advantages in both efficiency and accuracy.