Predefined-Time-Synchronized Whole-Body Control of Velocity-Controlled Mobile Manipulators With Unknown Kinematics

Abstract

The integration of wheeled mobile platforms and robotic manipulators presents challenges in deriving an accurate kinematic model of wheeled mobile manipulators (WMMs). This article investigates the whole-body kinematic control problem for velocity-controlled WMMs with unknown kinematics. We prove that conventional zeroing dynamics (ZD) models are not predefined-time-synchronized (PTS) stable and propose a predefined-time-synchronized ZD (PTS-ZD) model. This PTS-ZD model is subsequently utilized to address the whole-body inverse kinematics problem of WMMs. In addition, an iterative gradient dynamics (IGD) method is proposed to learn the unknown combined Jacobian matrix of WMMs. By integrating the PTS-ZD model with the IGD-based method, we propose a model-free algorithm to address the kinematic trajectory tracking challenge in WMMs with unknown kinematics. The proposed algorithm ensures that the tracking errors across different dimensions converge to zero synchronously within a predefined time. The model-free nature of the algorithm enables efficient and rapid adaptation to various WMM configurations. Finally, the effectiveness of the proposed method is validated through simulations and experiments on different WMMs.