论文部分内容阅读
为了实现机器人高精度运动轨迹的复现,需要把双目视觉采集到的离散的采样数据作为示教数据,由于机器人数学模型的实际参数与其名义值存在偏差,导致由视觉测量获得的机器人末端位姿逆解得到的关节变量直接复现时与原目标轨迹有一定的偏差。文中采用遗传算法对数据进行处理,将机器人运动学模型各参数的偏差都归结到关节变量上,建立位姿误差模型,求出目标函数最小时的关节变量修正值并补偿到控制器中,控制机械手完成高精度运动轨迹的复现。该方法避免了复杂的轨迹规划和传统的较为复杂的机器人标定补偿方法,提高了机器人复现精度,并能应用于任意复杂运动。在自行研制的6R串联机器人上对该方法进行了验证。
In order to reproduce the trajectory of the robot with high accuracy, the discrete sampling data collected by binocular vision needs to be taken as the teaching data. Since the actual parameters of the mathematical model of the robot deviate from their nominal values, the robot end bit obtained by the visual measurement The joint variables obtained from the inverse attitude solution have some deviations from the original target trajectory when they are directly reproduced. In this paper, the genetic algorithm is used to process the data, the deviation of each parameter of the robot kinematics model is attributed to the joint variable, the pose error model is established, the joint variable correction value when the objective function is minimum is calculated and compensated to the controller, Robot to complete high-precision trajectory of the recurrence. The method avoids the complicated trajectory planning and the traditional more complex calibration method of robot calibration, improves the accuracy of robot reproduction, and can be applied to any complex movement. The method was verified on a self-developed 6R tandem robot.