基于收缩法与数值导航函数的机器人路径规划

Robot path planning based on retraction method and numeric navigation function

将机器人的工作空间和状态空间离散化为分层的位图与网格，利用波前扩展技术完成自由空间到低维骨架网络的快速收缩，并将路径规划的收缩法与势场函数相结合，形成完整的数值导航函数的计算方法。给出了快速收缩与导航函数计算的核心算法，以校园建筑物轮廓图为障碍空间，建立了具有扩展功能的平坦越野环境下的路径规划器。所附的仿真实例表明，该方法规划速度快，较好地解决了全局规划与局部规划的有机衔接，并可有效处理高维空间中多自由度机器人的运动规划问题。

Robot's work space and configuration space is discretized as bitmaps and grids. A fast retraction from free space to a lower dimensional skeleton is performed by a wavefront expansion procedure. It combines the retraction method of motion planning with potential fields, and it forms a computational method for a complete numeric navigation function. The main algorithms of the quick retraction and navigation function computation are given. Taking building's contour of Tsinghua campus as obstacle space, a path planner with extended functionality is implemented under a smooth terrain environment. Simulation experiment included demonstrates that the method has the following capabilities: fast planning speed and effective coordination between global and local planning. It can also solve problems for robots with many more degrees of freedom in high dimensional space.