A Path Space Approach to Kinematic Path Planning

运动路径规划的路径空间方法

• 批准号: 9408874
• 负责人: John Wen
• 金额: $ 22.1万
• 依托单位: Rensselaer Polytechnic Institute
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1994
• 资助国家: 美国
• 起止时间: 1994-09-01 至 1998-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9408874&HistoricalAwards=false
• 关键词: Path; Space; Approach; Kinematic; Planning

The goal of this research is to produce a general and versatile path planner that can be applied to a wide range of complex situations including both multiple redundant robots and wheeled vehicles subject to obstacle avoidance and other constraints. Currently, path planning is a human--intensive endeavor with long and arduous teaching of the robot path by a human operator in a simulated work environment. Besides the cost and quality considerations of such practice, planning for certain complex systems, such as multifingered robot hands or tractors with multiple trailers, are beyond the capability of most humans. The approach of this research project is to first pose the path planning problem as a finite-time nonlinear control problem, and then transform it to a static root-finding problem in an augmented space which can be iteratively solved. The approach will be extended to address issues related to convergence, penalty function selection, optimality, and robustness and sensitivity with respect to the kinematic model. The algorithm will be systematically applied, in both simulation and experimentation n, to a number of fairly complex case studies involving both redundant manipulators and wheeled vehicles which have traditionally been treated as different cases.

本研究的目标是产生一个通用的、通用的路径规划器，可以应用于广泛的复杂情况，包括多个冗余机器人和受障碍物回避和其他约束的轮式车辆。目前，路径规划是一项人工密集的工作，需要人工操作员在模拟工作环境中对机器人进行长期而艰苦的路径教学。除了这种做法的成本和质量考虑外，对某些复杂系统的规划，如多指机械手或带有多个拖车的拖拉机，超出了大多数人的能力。本研究首先将路径规划问题作为有限时间非线性控制问题，然后将其转化为可迭代求解的增广空间中的静态寻根问题。该方法将被扩展到解决与收敛、惩罚函数选择、最优性以及相对于运动学模型的鲁棒性和敏感性相关的问题。该算法将在模拟和实验n中系统地应用于一些相当复杂的案例研究，涉及冗余操纵器和轮式车辆，这些案例传统上被视为不同的案例。