An Error Correction Method for Manipulator Flexibility Based on Elliptic Target Positioning

In the teleoperation of lunar surface sampling, a four axis robotic arm is adopted to collect lunar soils and transfer them to the sealed pot. Due to the flexibility of slender sampling arm and interval error of the joint, it is difficult to control the end-effector aligned with the mouth of the sealed pot without any feedback. In order to handle this problem, this paper proposes a method to calculate the position and normal orientation of sealed pot with respect to end-effector by using the elliptic features obtained from the monocular vision camera, based on which the angle of each arm joint is to be solved. Due to the rotational symmetry of round sealed pot, it is difficult to determine the rotations of the normal of the sealed pot. Nevertheless, considering special configuration constraints of the robotic arm, we can ignore the influence of undetermined rotation and solve the four joint angles uniquely. Thus, error of joint angles could be corrected and an effective control is exerted to the arm to guide the end effector gradually approaching the sealed pot. Finally, the effectiveness of the method is verified by experiments.