机器人视觉跟踪轨迹复现的新方法

A New Method of Reproducing a Trajectory Tracked by a Vision Guided Robot

基于视觉的机器人轨迹跟踪路径规划，需要把由视觉采集到的离散的采样数据复现成原曲线．本文提出用卡尔曼滤波的方法对数据进行处理，把滤波后的数据直接用于力矩规划，控制机械手完成复现的任务．这一方法避免了复杂的曲线拟合，减少了一个近似的环节，提高了控制精度并能不经任何改变地应用于任意的光滑曲线．本方法在自行研制的直接驱动机器人上进行了实验，取得了满意的结果．

This paper presents a new nethod for reproducing a trajectory tracked by a vision guidedrobot. Kalman filter is used to estimate data optimally which are read by a CCD camera mounted on the endeffector of the manipulator and are corruped by noise. After filtering,the data are sent directly into the controller to plan controlling torque. Comparing with previous methods,this way avoids doing complicated curvefitting and can be generalized to any kind of smooth curves including space curves without any modifications.Experiments have been implemented on our D. D. robot and give out satisfactory results.