Bayesian Optimization for Automatic Tuning of Torque-Level Nonlinear Model Predictive Control
This provides an automated tuning method for robotic control parameters, which is incremental but offers practical gains for robotics applications.
The paper tackles the problem of automatically tuning parameters for torque-based Nonlinear Model Predictive Control (nMPC) on a UR10e robot arm, using Bayesian Optimization with a digital twin to achieve a 41.9% improvement in tracking performance and 2.5% reduction in solve times in simulation, with 25.8% improvement in real-world experiments.
This paper presents an auto-tuning framework for torque-based Nonlinear Model Predictive Control (nMPC), where the MPC serves as a real-time controller for optimal joint torque commands. The MPC parameters, including cost function weights and low-level controller gains, are optimized using high-dimensional Bayesian Optimization (BO) techniques, specifically Sparse Axis-Aligned Subspace (SAASBO) with a digital twin (DT) to achieve precise end-effector trajectory real-time tracking on an UR10e robot arm. The simulation model allows efficient exploration of the high-dimensional parameter space, and it ensures safe transfer to hardware. Our simulation results demonstrate significant improvements in tracking performance (+41.9%) and reduction in solve times (-2.5%) compared to manually-tuned parameters. Moreover, experimental validation on the real robot follows the trend (with a +25.8% improvement), emphasizing the importance of digital twin-enabled automated parameter optimization for robotic operations.