Event‐Driven Model Predictive Control With Sliding Mode Observer for Omnidirectional Demolition Robots

This article addresses the trajectory tracking problem for a fully omnidirectional demolition robot (FM‐OMR) equipped with four Mecanum wheels under longitudinal wheel slippage and unknown disturbances. It proposes an event‐triggered model predictive control (EMPC) framework and introduces an adaptive exponential reach law sliding mode observer (AERL‐SMO) for online disturbance estimation. The AERL‐SMO suppresses the chattering inherent to classical sliding‐mode observers by using tanh‐based smoothing and a boundary‐layer saturation function, and it accelerates convergence for large errors through an adaptive exponential gain. The EMPC employs an event‐triggered strategy: the MPC is re‐solved only when the error energy exceeds a threshold, which significantly reduces the number of online QP solves. Simulation validation demonstrates that under identical constraints and disturbance conditions, the proposed method reduces MPC solution iterations from 1257 to 579 (a 53.85% decrease) while substantially lowering the mean absolute error (MAE) and disturbance estimation error. The article also presents stability analysis (based on Lyapunov) and parameter sensitivity discussions for the algorithm. This method is suitable for omnidirectional mobile robot platforms with limited computational resources that require resistance to wheel slippage and complex disturbances.