Data-driven Acceleration of MPC with Guarantees

2025-11-17

AI Summary

To address the high online computational latency of Model Predictive Control (MPC), which hinders real-time deployment, this paper proposes a data-driven, nonparametric acceleration framework. The method constructs an upper bound on the optimal cost from an offline MPC solution dataset and employs a greedy lookup strategy to replace online optimization with table-based inference. It establishes, for the first time, explicit theoretical guarantees on recursive feasibility and bounded suboptimality, quantifying the trade-off between dataset size and performance bounds. Experimental results demonstrate that the proposed approach accelerates MPC by 100–1000× compared to conventional implementations, while incurring only negligible performance degradation. Crucially, control quality—measured in terms of stability, constraint satisfaction, and closed-loop performance—is rigorously preserved. This work thus introduces a new paradigm for low-latency, real-time control grounded in data-driven approximation with provable guarantees.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Optimization
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Model Predictive Control (MPC) is a powerful framework for optimal control but can be too slow for low-latency applications. We present a data-driven framework to accelerate MPC by replacing online optimization with a nonparametric policy constructed from offline MPC solutions. Our policy is greedy with respect to a constructed upper bound on the optimal cost-to-go, and can be implemented as a nonparametric lookup rule that is orders of magnitude faster than solving MPC online. Our analysis shows that under sufficient coverage condition of the offline data, the policy is recursively feasible and admits provable, bounded optimality gap. These conditions establish an explicit trade-off between the amount of data collected and the tightness of the bounds. Our experiments show that this policy is between 100 and 1000 times faster than standard MPC, with only a modest hit to optimality, showing potential for real-time control tasks.

Problem

• Accelerating MPC for low-latency applications
• Replacing online optimization with offline data-driven policy
• Ensuring recursive feasibility and bounded optimality gap

Innovation

• Data-driven MPC acceleration via offline policy
• Nonparametric lookup rule replaces online optimization
• Guarantees recursive feasibility and bounded optimality gap