运动规划Motion Planning
Motion Planning
给机器人算出一条从起点到目标、躲开障碍又能执行的动作轨迹。Compute an executable action trajectory from start to goal that avoids obstacles and respects constraints.
运动规划在机器人的位形空间里搜索一条满足约束的路径：避开障碍、不超关节限位，再从路径生成随时间执行的动作序列。
常用做法是基于采样的规划器，如 PRM 与 RRT，靠随机采样加连接快速在高维空间找到可行解，OMPL 汇集了这类算法。
MoveIt 把规划、碰撞检查与执行串成机器人栈里的现成管线，是 ROS 上机械臂运动规划的常用入口。
Motion planning searches the robot configuration space for a path satisfying constraints: avoid obstacles, stay within joint limits, then generate a time-parameterised action sequence from the path.
Sampling-based planners such as PRM and RRT are common, using random sampling and connection to find feasible solutions quickly in high-dimensional spaces; OMPL collects such algorithms.
MoveIt connects planning, collision checking and execution into a ready pipeline in the robot stack, and is a common entry point for arm motion planning on ROS.
机器人「怎么把手伸过去而不撞到杯子」靠的就是它。知道规划是在算可行轨迹而非凭感觉动，你就能分清一段演示是预先规划好的，还是实时学出来的策略在控制。How a robot reaches out without hitting the cup depends on this. Knowing planning computes feasible trajectories rather than moving by feel helps distinguish pre-planned demonstrations from real-time learned policy control.