Robust torque-observed control with safe input–output constraints for hydraulic in-wheel drive systems in mobile robots

Abstract

Hydraulic-powered in-wheel drive (IWD) mechanisms enhance the maneuverability, traction, and maintenance efficiency of heavy-duty wheeled mobile robots (HWMRs) by enabling independent operation of each wheel. Sufficient motion in such HWMR systems relies on a multi-stage power transmission mechanism that integrates control valves, hydraulic motors, gearboxes, and, ultimately, nonlinear ground-interaction wheel dynamics on rough terrain. Deviations in each stage of these independently operated wheel systems—arising from modeling uncertainties and disturbances such as wheel slippage and uneven torque distribution on rough terrain—can disrupt motion balance between wheels and further amplify deviations. This can lead the robot to deviate from its course, oscillate, or lose traction, ultimately resulting in overall instability, which may pose a risk to the heavy-weight robot's surrounding environment. To develop a synchronous control strategy for distributed HWMR systems to mitigate such challenges in uncertain environments, this paper proposes a novel robust torque-observer-based valve control (RTOVC) framework for IWD-actuated wheels, guaranteeing robustness and uniformly exponential stability of the entire system. As a foundation for this approach, a robust torque observer network based on an adaptive barrier Lyapunov function (BLF) is designed to obtain the required wheel/motor torques, ensuring that the actual velocities of IWD-actuated wheels align with the reference values in motion dynamic frames in the presence of wheel slippages. It eliminates the closed-loop dependency on fault-prone torque or pressure sensors in hydraulic actuation mechanisms. Building on this, an additional adaptive BLF-based control network in the valve-actuated hydraulic mechanism is employed to regulate fluid flow, generating the required torque in the first network for each wheel under system uncertainties. The RTOVC framework reduces fault risks in HWMRs by constraining key input–output signals—such as valve control signals, actual wheel velocities, tracking errors, and required motor/wheel torques—within logarithmic BLFs, ensuring safe operation. A comprehensive experimental analysis on a 6,500-kg hydraulic-powered IWD-actuated HWMR operating on rough terrain, where failures may arise due to severe slipping conditions and hydraulic system uncertainties, confirms the RTOVC's robust performance compared with two other state-of-the-art control strategies.