In this paper, a model-free gravity compensation algorithm for manipulators based on L_1-norm convex optimization is proposed. The method does not depend on the structure and parameters of the robotic arm. In contrast, this method optimizes the solution of gravity compensation parameters based on the sampled data. Theoretical analysis shows that the reconstruction of the gravity compensation parameters of the manipulator is realized based on the joint angle and joint torque training data. Experiments show that the gravity compensation results can restore the joint torque of the test set well, which proves the effectiveness of the method.
展开▼