Real-Time Delta Laguerre Model Predictive Control for a Flexible Robot with Nonlinear Inverse Model Feedforward

Kurzfassung

This paper presents a Delta Laguerre Model Predictive Control (MPC) framework for real-time industrial robot axis control operating at 2000 Hz. The controller employs a delta formulation with Laguerre functions to optimize incremental control efforts while integrating a nonlinear inverse dynamics feedforward model. To capture joint flexibility, a discrete two-mass-spring-damper model is derived that adapts to changing robot configurations through real-time parameter scheduling. A fixed-size iterative Riccati solver is implemented to ensure deterministic execution within the 0.5 ms sampling interval, effectively handling system ill-conditioning caused by dead-time augmentation. The controller is implemented in Modelica and exported as a Functional Mock-up Unit (FMU) for deployment. Experimental results on a KUKA KR16 industrial robot with a advanced control setup demonstrate improved tracking performance and constraint handling compared to standard PPI control. Furthermore, a Monte Carlo analysis verifies the system's robustness against significant parameter variations.