Graph Neural Model Predictive Control for High-Dimensional Systems

TL;DR

This paper introduces a GNN-based control framework for high-dimensional systems like soft robots, enabling real-time, accurate control by combining structure-exploiting MPC with efficient graph-based modeling.

Contribution

It presents a novel integration of Graph Neural Networks with Model Predictive Control, optimized for high-dimensional systems with linear complexity and GPU acceleration.

Findings

Scales to systems with up to 1,000 nodes at 100 Hz in real-time.

Achieves sub-centimeter accuracy in soft robotic control.

Outperforms baseline methods by 63.6% in tracking performance.

Abstract

The control of high-dimensional systems, such as soft robots, requires models that faithfully capture complex dynamics while remaining computationally tractable. This work presents a framework that integrates Graph Neural Network (GNN)-based dynamics models with structure-exploiting Model Predictive Control to enable real-time control of high-dimensional systems. By representing the system as a graph with localized interactions, the GNN preserves sparsity, while a tailored condensing algorithm eliminates state variables from the control problem, ensuring efficient computation. The complexity of our condensing algorithm scales linearly with the number of system nodes, and leverages Graphics Processing Unit (GPU) parallelization to achieve real-time performance. The proposed approach is validated in simulation and experimentally on a physical soft robotic trunk. Results show that our method scales to systems with up to 1,000 nodes at 100 Hz in closed-loop, and demonstrates real-time reference tracking on hardware with sub-centimeter accuracy, outperforming baselines by 63.6%. Finally, we show the capability of our method to achieve effective full-body obstacle avoidance.