论文部分内容阅读
提出一种视觉定位方法对影响老年服务机器人的视觉定位精度的摄像机标定、手眼系统标定以及三维重建3个部分进行优化.首先对多幅标定图像的结果进行最小二乘处理,剔除大权重的误差样本以减小标定的统计误差;通过建立过渡空间坐标系的方法来对手眼系统进行标定,减少多次计算转移矩阵所造成的累积误差;利用最小二乘拟合精确确定目标的匹配点,减小三维重建所带来的误差.实验结果表明:使用该方法使机器人获取目标空间位置的误差小于2mm,满足老年服务机器人路径规划及使用手爪抓取物体的需要.
A visual positioning method is proposed to optimize the camera calibration, hand-eye system calibration and 3D reconstruction that affect the visual positioning accuracy of the elderly service robot.First, the results of multiple calibration images are processed by least squares, and the error of large weights Sample to reduce the statistical error of calibration; through the establishment of a transition space coordinate system to calibrate the hand-eye system to reduce the cumulative error caused by multiple transfer matrix calculation; using least-squares fitting to accurately determine the target matching point, minus The error caused by small three-dimensional reconstruction.The experimental results show that the error of the robot to obtain the target spatial position is less than 2mm, which meets the needs of path planning of old service robots and grasping objects by using hand claws.