An inertial neural network approach for loco-manipulation trajectory tracking of mobile robot with redundant manipulator

This paper proposes a novel constrained optimization model to address the loco-manipulation problem of mobile robot with redundant manipulator for trajectory tracking. To alleviate the accumulative error of the end-effector's position, a new control law is designed to eliminate the negative effect from the deviation of the initial position, leading to better performance than existing ones. To deal with the locomotion constraints in the loco-manipulation problem, the optimization model is converted to an augmented Lagrangian primal-dual problem. Furthermore, an inertial neural network approach is used to solve the problem and the corresponding Lyapunov proof guarantees the convergence of variables. The numerical simulations show that the proposed approach is more suitable for application since the model is more effective and the algorithm has better convergence rate.