Learning Nonlinear Systems In-Context: From Synthetic Data to Real-World Motor Control

2026-02-06

AI Summary

This work addresses the limitations of conventional PI control and physics-based feedforward methods in effectively regulating motor systems under nonlinear and complex loading conditions. The authors propose a novel Transformer-based architecture that leverages signal-behavior decoupled modeling and is pretrained on extensive synthetic data encompassing both linear and nonlinear dynamical systems. Notably, this approach introduces in-context learning (ICL)—a mechanism borrowed from large language models—into motor control for the first time. The framework enables single-shot contextual inference and few-shot fine-tuning, substantially bridging the sim-to-real gap. Experimental results demonstrate that, across diverse motor load configurations, the method generates high-precision feedforward signals using only a handful of examples, significantly outperforming traditional PI controllers and physics-based baselines.

Technology Category

• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Intelligent Robots: Behavior Learning & Control
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

LLMs have shown strong in-context learning (ICL) abilities, but have not yet been extended to signal processing systems. Inspired by their design, we have proposed for the first time ICL using transformer models applicable to motor feedforward control, a critical task where classical PI and physics-based methods struggle with nonlinearities and complex load conditions. We propose a transformer based model architecture that separates signal representation from system behavior, enabling both few-shot finetuning and one-shot ICL. Pretrained on a large corpus of synthetic linear and nonlinear systems, the model learns to generalize to unseen system dynamics of real-world motors only with a handful of examples. In experiments, our approach generalizes across multiple motor load configurations, transforms untuned examples into accurate feedforward predictions, and outperforms PI controllers and physics-based feedforward baselines. These results demonstrate that ICL can bridge synthetic pretraining and real-world adaptability, opening new directions for data efficient control of physical systems.

Problem

• nonlinear systems
• in-context learning
• motor control
• feedforward control
• system identification

Innovation

• In-Context Learning
• Transformer
• Motor Control
• Nonlinear Systems
• Synthetic-to-Real Generalization