arXiv · 2611.15934
Model Predictive Control on Analogue Hardware: A Dual-flow Approach
Abstract
Current Model Predictive Control (MPC) implementations are predominantly digital and, despite offering high precision and flexibility, often face challenges with ultra-low latency and hardware certification for safety-critical applications. We present an approach for solving the Quadratic Program (QP) of MPC using a continuous-time neural ordinary differential equation in the dual space. We obtain the solution to these QPs by mapping the differential equation's equilibrium point back to the primal space via a nonlinear transformation. The underlying dual-flow dynamics operate natively on analogue hardware. We validate this approach by demonstrating the hardware implementation of linear MPC on "The Analog Thing" analogue computer. Finally, we extend the framework to nonlinear MPC with successive linearisation, broadening the applicability beyond linear systems.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Nikilesh Ramesh, Ross Drummond, Pablo R Baldivieso-Monasterios, Morgan Jones, Yuanbo Nie. 2026-11-11. Model Predictive Control on Analogue Hardware: A Dual-flow Approach. https://searcharxiv.com/abs/2611.15934
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.