Data-Driven Approaches to Real-Time Constrained Tracking Control of Redundant Manipulators

This study proposes two contrasting neural control architectures that rely on the use of input–output information to compensate for the lack of knowledge about the robot model to achieve the prescribed tracking performance in the presence of joint angle and velocity constraints. First, an observer–controller zeroing neural network framework is formulated, which combines online estimation of the unknown model’s Jacobian with a trajectory tracking controller that implements joint angle and velocity constraints via a nonlinear map. Further, prescribed performance constraints are embedded within this architecture to achieve the desired transient and steady-state performance. This allows for the implementation of state and input constraints with the added benefits of superior transient and steady-state performance. Alternatively, a Koopman operator theoretic framework is also considered for real-time control of redundant manipulators subject to input constraints. An autoencoder-based neural architecture is employed to discover the bilinear Koopman model for manipulator dynamics in joint space using input–output data. This architecture is subsequently integrated with a feedforward neural network that maps the joint coordinates to end-effector Cartesian coordinates, which allows for the learning of highly accurate models with a significantly lower number of observable states compared to the previous studies. The learning architecture is then coupled with a zeroing dynamics neural controller, which offers a computationally inexpensive alternative to state-of-the-art nonlinear model predictive controllers whose computational burden might render real-time control infeasible. The low-dimensional learned model, when combined with the computationally inexpensive control policy, facilitates real-time control applications with improved tracking accuracy. Simulation studies of trajectory tracking are used to verify the efficacy of the proposed schemes.