Experimental Study of Model-based Torque Control for 6-DOF Manipulator Robot

The torque control algorithm for a robot computes joint torque commands using inverse dynamic equations, enabling the robot to follow reference trajectories with high precision. By leveraging whole-body dynamic information, model-based control allows robots to perform well in collaborative tasks and remain robust against disturbances caused by predicted actions. This paper describes the entire process of constructing a model-based controller, considering practical factors such as algorithm implementation and servo controller compensation. To implement the model-based control algorithm, kinematic parameters are extracted from the 3D CAD design. Since various interfaces for inverse dynamics calculations use different coordinate systems, the coordinate system must be adjusted appropriately for each context. Then, parameters such as inertia tensors are transformed into the relevant coordinate system. Subsequently, the rigid-body dynamics are formulated, and the control algorithm is derived. To achieve more precise feedback control, each joint torque sensor is calibrated for torque control. As a final step, a model-based controller is implemented in a commercial manipulator robot, and its implementation is substantiated through a comparison with independent joint control.