未知环境中移动机器人探索式路径规划方法研究

In unknown environments, reaching the destination in an optimal way is a practical problem confronting mobile robot. Path planning is the core issue of searching optimal path, while getting required environment information is the critical problem of path planning. So, an optimal path in unknown environment should consider more factors than it needs in known one, such as the influence of mapping, localization, and the environmental exploration task should be covered as well. .By considering the influence among path, mapping and localization at the same time, an evaluating model based on entropy is first proposed in this project. With the study of the distribution law of path, an evaluating model for the unknown part of the environment will be established, which would be segmented according to some key characters recognized from the environmental map. So, the evaluating model of the optimal path can then be established based on the two models. .Then, the global optimal path is treated as a changing state variable which is affected by the motion of the robot, the localization of the robot and the map of the environment. The path planning is also transferred to an estimating problem. For this new problem, an exploratory path planning method based on particle filter would be put forward. Finally, the environmental exploration could be guided by the path planning result, and the path planning could be carried out based on the result of exploration..Generally, this project proposed a new approach to improve the performance of mobile robot in unknown environment and it will advance relevant research of mobile robot greatly.

未知环境中，如何以最优方式到达目标是移动机器人面临的现实问题。路径规划是解决该问题的核心，而获取足够的环境信息则是进行有效路径规划的关键。因此，未知环境中的最优路径应综合考虑定位、绘图的影响，考虑探索环境地图未知部分的需求。.本项目首先研究路径与定位、绘图间的影响机理，建立基于熵的定位、绘图影响评价模型，探索路径分布规律，基于环境特征建立对地图未知部分的分区评价模型，从而形成对最优路径的评价模型，解决未知环境中最优路径的评价标准问题。其次，将未知环境中最优路径视为受地图、定位、运动影响而变化的状态量，基于粒子滤波提出一种探索式路径规划方法，将路径规划问题转化为对状态的最优估计问题，从而以路径规划指导环境探索，并反馈支持路径规划，从而解决未知环境中难以获取足够环境信息的问题。.项目成果将为解决未知环境中移动机器人导航问题提供一种新思路，并为相关领域研究提供有益参考。

未知环境中，如何以最优方式到达目标是移动机器人面临的现实问题。路径规划是解决该问题的核心，获取足够的环境信息则是进行有效路径规划的关键。因此，未知环境中的最优路径应综合考虑定位、绘图的影响，考虑探索环境地图未知部分的需求。. 项目首先研究路径与定位、绘图间的影响机理，建立基于深度学习的可定位性评估方法，建立绘图影响评价模型，探索路径分布规律，基于环境特征建立对地图未知部分的分区评价模型。研究定位误差对路径的影响机理，形成对最优路径的新的评价模型，解决未知环境中最优路径的评价标准问题。其次，将未知环境中最优路径视为受地图、定位、运动影响而变化的状态量，基于粒子滤波提出一种探索式路径规划方法，将路径规划问题转化为对状态的最优估计问题，从而以路径规划指导环境探索，并反馈支持路径规划，从而解决未知环境中难以获取足够环境信息的问题。. 项目成果将为解决未知环境中移动机器人导航问题提供一种新思路，并为相关领域研究提供有益参考。