Graph Neural Model Predictive Control for High-Dimensional Systems
2026-02-19 • Robotics
Robotics
AI summaryⓘ
The authors developed a method to control complex systems like soft robots by combining special neural networks that model parts of the system as a graph with a control technique that takes advantage of this structure. Their approach makes calculating control signals faster and keeps things simple by removing unnecessary variables, scaling well even for very large systems. They tested it both in simulations and on a real soft robotic trunk, showing it works accurately and quickly, better than previous methods. Additionally, their method can handle full-body movements to avoid obstacles effectively.
Graph Neural NetworksModel Predictive ControlSoft RobotsHigh-dimensional SystemsSparsityParallelizationGPU ComputingReal-time ControlReference TrackingObstacle Avoidance
Authors
Patrick Benito Eberhard, Luis Pabon, Daniele Gammelli, Hugo Buurmeijer, Amon Lahr, Mark Leone, Andrea Carron, Marco Pavone
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.