视觉引导的机器人轨迹跟踪路径规划的一种新方法

详细信息

A NEW APPROACH OF PATH PLANNING FOR THE ROBOT TRAJECTORY TRACK BASED ON VISION

摘要: 在基于视觉的机器人轨迹跟踪路径规划中，必须将离散的采样数据复现成原曲线．本文提出一种利用曲线上任意点的切线与水平线的夹角和弧长之间的关系的方法，用圆弧段和直线段来近似拟合二维图线，克服了通常曲线拟合方法不保证高阶导数收敛以及对数据噪声不理想的弱点，实现了曲线的较精确复现，并在自制的直接驱动机器人上进行了实验，取得了满意的效果．
- 路径规划 /
- S-θ图 /
- 分段 /
- 滤波
Abstract: During the path planning of the robot trajectory track based on vision,it's necessary to reproduce the discrete sampled data into the original curve.The paper provides a new approach of fitting the are and straight lines using the relationship of the arc length and the angle between the tangent of the curve on the point and the horizon. The common two dimension curve fitting methods can't converge the high order derivatives and conquer the data noises. Our method can realize the exact reproduction of the curve.Experiment has been made on our Direct Drive robot.The result is quite good.
- Trajectory plan /
- S-θcurve /
- segment /
- filter

[1]	苏步青.微分几何学.复旦大学出版社,1985.
[2]	蒋平.机器人手控制方法与并行实现研究.西安交通大学博士论文,1992.
[3]	陈辉堂.数字控制系统.西安交通大学出版社,1986.