Sym2Real: Symbolic Dynamics with Residual Learning for Data-Efficient Adaptive Control

2025-09-18

AI Summary

This work addresses the low data efficiency, expert-dependent hyperparameter tuning, and reliance on high-fidelity simulation inherent in low-level adaptive controllers deployed on real robotic systems. We propose a robust sim-to-real transfer framework requiring only ~10 real-world trajectories. Methodologically, it integrates low-fidelity simulation modeling, directed residual learning, and symbolic regression—the first application of symbolic regression to real-robot control—where physics-informed constraints enhance noise robustness and generalization, while shared physical structure enables cross-environment adaptation. Technically, the framework unifies system identification and residual compensation to ensure safe and reliable transfer. Experiments demonstrate stable adaptation across six out-of-distribution simulation scenarios and achieve high-performance control on five diverse real platforms—including quadcopters and autonomous race cars—substantially reducing real-data requirements.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We present Sym2Real, a fully data-driven framework that provides a principled way to train low-level adaptive controllers in a highly data-efficient manner. Using only about 10 trajectories, we achieve robust control of both a quadrotor and a racecar in the real world, without expert knowledge or simulation tuning. Our approach achieves this data efficiency by bringing symbolic regression to real-world robotics while addressing key challenges that prevent its direct application, including noise sensitivity and model degradation that lead to unsafe control. Our key observation is that the underlying physics is often shared for a system regardless of internal or external changes. Hence, we strategically combine low-fidelity simulation data with targeted real-world residual learning. Through experimental validation on quadrotor and racecar platforms, we demonstrate consistent data-efficient adaptation across six out-of-distribution sim2sim scenarios and successful sim2real transfer across five real-world conditions. More information and videos can be found at at http://generalroboticslab.com/Sym2Real

Problem

• Data-efficient adaptive control for robotics
• Overcoming noise sensitivity in symbolic regression
• Sim2real transfer with residual learning

Innovation

• Symbolic regression with residual learning
• Low-fidelity simulation plus real-world data
• Data-efficient adaptive control framework