An Optimization-based direct teaching method for manipulator with tool's gravity compensation

In this paper, a direct teaching method for operator is proposed to replace teaching pendant. The direct teaching algorithm is inspired by impedance control, which can be used to generate joint position commands from force input. Force signal is filtered with first order inertial filter, and tool's gravity is removed with zero drift for taking more precise control into consideration. In order to achieve such as abiding by physical joint limits and avoiding high frequency oscillation, we formulate the whole inverse kinematics as an optimization problem. The application and experiment are conducted on UR10 robot with a 6-axis force sensor attached to the end-effector.