Sparse interpretable feedback control improves nonlinear system performance
SD-DPC: Sparse Dictionary Differentiable Predictive Control
Machine Learning
Summary
Controlling complex systems like robots or vehicles can be very complicated and require a lot of computing power. The authors developed a method to learn simple, easy-to-understand control rules from data that can predict how the system will behave and adjust actions accordingly. Their approach uses a combination of mathematical functions, selecting only a few important ones to keep the control policy small and efficient. Tests on standard control tasks show their method can perform better than similar approaches while using less memory and computing power.
What this means in practice
- •For robotics engineers: Implement compact predictive controllers that maintain constraints while requiring less computational resources for nonlinear robot systems.
- •For automotive control teams: Design efficient feedback control policies for nonlinear vehicle dynamics that reduce onboard memory and processing needs.
Authors
Ali Reza Daneshvar Garmroodi, Jan Drgoňa
Abstract
We present sparse dictionary differentiable predictive control (SD-DPC), a framework for learning sparse, interpretable feedback policies for nonlinear systems from data. A prediction model is first identified by rollout-based sparse identification of nonlinear dynamics (SINDy), building on gradient-based and multistep formulations. The policy is then parameterized as a sparse combination of dictionary functions and trained by differentiating a constrained finite-horizon predictive-control objective through this model, so that its terms are selected by closed-loop performance rather than by imitating a previously trained controller. The result is an explicit feedback law with only a handful of terms. Across three benchmark control problems, SD-DPC satisfies the constraints in all test scenarios, outperforms a policy distilled onto the same terms by up to an order of magnitude, and requires orders of magnitude less memory and online computation than an optimization benchmark, while admitting explicit sensitivity bounds.