广域动态的野外环境中移动机器人六维全局定位方法的研究

Global localization has long been considered as one of the most important but also challenging problems for mobile robots operating in the large-scale and dynamic field environment. Current studies of global localization in literature are mainly based on the Bayesian filtering technique which can provide an elegant statistical framework for the uncertainty management and multisensory fusion. However the majority of implementations of Bayesian-filters for global localization obey the same update rules in such a location driven sense that they guess the robot location first and then adjust the guess by incorporating the current observation data. This leads to some problematic consequence that the system suffers from great computational load in large application area and it cannot recover from localization failure. Thus the majority of conventional global localization systems cannot be applied in the large-scale and dynamic field environment. Thus this project breaks out the conventional update rules of Bayes filters and proposes a new approach – the observation driven Bayes filters (OD-BF). As the name implies, OD-BFs estimate the robot state just according to the most recent observations and then adjust the estimate by incorporating the dead reckoning information. We further implement an observation driven Bayes filter for globally estimating the robot pose in the large-scale and dynamic field environment. This global localization system features in an effective feature matching framework which has a high robustness against uncertainties from sensor occlusions and extraneous observation in a highly dynamic field environment. Additionally our project implements a data fusion system which is able to deal with global initialization and have the ability of quick recovery from a localization failure. We will carry out sufficient practical experiments to determine both advantages and disadvantages of our proposed global localization system.

定位问题是移动机器人研究领域中一个基础且重要的问题。近年来，移动机器人的应用逐步拓展到了很多广域动态的野外环境中，野外环境下的移动机器人定位问题凸现其研究价值。本课题将针对广域动态的野外环境，研究观测驱动型全局定位方法中的地图匹配问题和数据融合问题，进而提出一种新的观测驱动型全局定位系统，实现在广域动态的野外环境中移动机器人六维全局定位。课题研究在地形起伏的野外环境中三维环境信息的表征、提取和组织方法；研究在存在动态干扰物和观测噪声的前提下，高效鲁棒的三维地图匹配技术；进而研究基于观测驱动型贝叶斯滤波器的数据融合方法，其将有效处理野外环境中的各种不确定性，并在"定位失败"的前提下快速重新定位。课题将在真实的广域动态野外环境中开展全局定位的实验。本课题研究对移动机器人野外环境中的定位理论的发展以及拓展移动机器人在野外环境中的应用具有积极意义。

定位问题是移动机器人研究领域中一个基础且重要的问题。近年来，移动机器人的应用逐步拓展到了很多广域动态的野外环境中，野外环境下的移动机器人定位问题凸现其研究价值。本课题针对广域动态的野外环境，研究并提出提出一种新的观测驱动型全局定位系统，实现在广域动态的野外环境中移动机器人六维全局定位。首先，本课题设计和实现了一种观测驱动型贝叶斯滤波方法，可以实现在野外大型环境下有效、可靠地数据融合工作，还可以有效解决机器人“定位失败”后的“重定位”问题；增加在野外环境中定位系统的鲁棒性；为解决全局定位问题中地图匹配的高计算复杂度问题，本课题提出了一种基于渐进型的匹配思路的高效、鲁棒的地图匹配算法。渐进型匹配算法分为两步：1）粗匹配过程，其核心基于“最小特征匹配”算法；本课题设计的基于“最少特征匹配对” 搜索的地图匹配算法解决了完整搜索“特征匹配对空间”所需的指数级算法复杂度的问题，可以在常用的计算机平台上做到实时计算。2）基于ICP的位姿假设修正方法，解决了粗匹配带来的误差问题。最后，本课题提出了野外环境特征信息的表征和提取方法。提出了针对全局地图的八叉树表征方发以及针对局部地图的“关键特征代表点选择算法”；从而实现了地图匹配过程中“渐进型的匹配”过程，既保证了匹配的精度和鲁棒性，也能解决传统地图匹配搜索的高计算复杂度。本课题搭建了轮式移动机器人实验平台，并在真实的野外环境中开展了充分的实验,验证了所实现的定位系统的全局定位性能。本课题累计发表SCI论文10篇。