Approximation-Free Robust Tracking Control of Unknown Redundant Manipulators With Prescribed Performance and Input Constraints

This article proposes a novel neural control architecture that employs input-output information to compensate for the lack of knowledge about the robot model to achieve prescribed tracking performance in the presence of joint constraints. To this end, an observer-controller zeroing neural network framework is formulated that 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 desired transient and steady-state performance along with added robustness to chattering. Hence, in comparison to prior studies, the proposed scheme facilitates a more robust control architecture with the added benefits of more stringent application of the input constraints and superior transient and steady-state performance. Simulation and experimental studies of trajectory tracking, including comparisons with leading alternative designs, are used to verify the efficacy and superior performance of the proposed scheme.

I. Introduction

Trajectory tracking for redundant manipulators represents a preeminent and fundamental problem within the field of robotics. It is a subset of a wider class of problems pertaining to redundancy resolution, which refers to the computation of the joint poses required to obtain the desired end-effector positions for a redundant manipulator [1]. In particular, redundancy resolution relies on determining the inverse kinematic map from the end-effector position to the joint pose. However, due to the nonlinear nature of the forward map, the process of finding the required inverse map can prove to be quite challenging. Hence, to avoid the computational complexity, redundancy resolution is usually solved at the velocity level where the problem can be represented as a time-varying underdetermined set of linear equations [2]. The underdetermined nature of the system allows for the inclusion of additional functionalities into the system, such as state and input constraints, and obstacle avoidance [3], [4].