基于全局动态非结构化人居环境感知的仿人机器人足迹规划研究

项目名称： 基于全局动态非结构化人居环境感知的仿人机器人足迹规划研究

项目编号： No.51305436

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 机械、仪表工业

项目作者： 夏泽洋

作者单位： 中国科学院深圳先进技术研究院

项目金额： 26万元

中文摘要： 实现在复杂非结构化的人居环境中的运动是仿人机器人学研究的重要目标。已有的仿人机器人运动规划方法通常在无环境约束或局部环境约束下进行，难以适用。基于采样的足迹规划是考虑全局环境信息的仿人机器人路径规划，是解决上述问题的有效方法。本课题组在前期研究中已经实现了在静态环境中的足迹规划，本课题拟在此基础上开展动态复杂非结构化环境中的足迹规划研究。主要研究内容有：（1）建立用于人居环境感知的全景视觉系统，构建同步三维数字地图，作为足迹规划器的实时输入；(2)建立用于动态环境的实时滚动足迹规划方法,设计包含基于在线环境评估进行多类足迹规划器优化切换的机制和规划异常处理机制的混杂足迹规划器。(3) 进行仿真实验，并建立仿人机器人物理平台，在典型动态人居环境中对上述系统和关键理论与技术进行验证。本研究的目标是实现仿人机器人在动态人居环境中的运动，为仿人机器人在不久的将来进入普通百姓家庭服务奠定基础。

中文关键词： 仿人机器人；双足步行；足迹规划；环境感知；全景视觉

英文摘要： To realize biped walking in complicated non-structured human-living environments is an important objective of humanoid robotics research. Existing biped walking approaches are not feasible in the above environments because they consider only local of or even no environment constraints. Sampling-based footstep planning is a path planning approach for humanoid robots considering global environment information, thus is an effective method to resolve the above problem. Previously, our group has realized footstep planning in static environments. We are now proposing to study the footstep planning in complicated non-structed dynamic environment. The following tasks will be conducted in our study. (1) An omni vision system will be constructed to obtain the simultaneous three-dimensional digital map of the dynamic environment, which will be used as input of the footstep planner. (2) Footstep planning method for dynamic environments will be studied. A hybrid footstep planner including an online swtich mechanism for selecting apporiate planner in different environments and exception handler will be constructed. (3) Besides simulation validation, a physical humanoid robot will be designed to test and validate the above system and methods in typical dynamic human-living environments. The Objective of our study is to reali

英文关键词： Humanoid robot；Biped walking；Footstep planning；Environment perception；Omni vision