摘要
针对挖掘机机器人化作业控制系统中,测量操作臂姿态的旋转编码器等接触式传感器易因碰撞而损坏,传统的非接触式测量系统测量范围小,响应速度慢等不足,建立了一种挖掘机操作臂姿态非接触式实时测量系统。该系统基于机器视觉和三组人工靶标,操作臂的姿态用人工靶标中的三个X角点所在直线方向表示,并以此为基础,提出X角点分组跟踪策略,将各组靶标中X角点的检测范围限定在某一特定矩形区域内,应用X角点的间距作为约束条件完成人工靶标图像识别,实时计算得到操作臂姿态参数。动态测量试验表明:动臂和铲斗姿态测量误差分别在±0.5°、±1°内;X角点分组跟踪策略缩减了80%的图像检测面积,测量平均耗时仅63ms/帧,满足挖掘机实时控制要求。
The contacting sensors used by the contacting measuring system would be destroyed in the excavatorautomotive work process,also,the present non-contact measuring systems based on computer vision cannot be applied to the measurement for the attitude of excavator's manipulators. A novel artificial target was developed which possess 3 collinear X-corner and the manipulators' attitude was presented by the direction of the line. The method of tracking X-corners in group was proposed to narrow the X-corner detecting area,and artificial targets' X-corners would be detected and matched according to its' actual range,then manipulators' attitude could be obtained. On the basis of experiments,the measurement error range of boom's[-0.5°,0.5°],and bucket's[-1°,1°];the detecting area of an image is reduced 80% compared to the picture by the implement of the method of tracking X-cornets in groups,measurement takes 63 ms per frame on the average.
出处
《机械设计与制造》
北大核心
2018年第3期105-108,共4页
Machinery Design & Manufacture
基金
国家重点实验室开放基金资助项目(GZKF-201308)
关键词
挖掘机
非接触式测量
人工靶标
分组跟踪
实时
Excavator
Non-Contact Measuring
Artificial Target
Tracking in Group
Real-Time
