Meta-Learning Online Dynamics Model Adaptation in Off-Road Autonomous Driving

2025-04-23

AI Summary

High-speed off-road autonomous driving faces significant challenges in unknown, unstructured terrains due to poor generalizability of vehicle dynamics models and difficulty in modeling terrain–vehicle interactions. To address these issues, this paper proposes an online adaptive framework that synergistically integrates meta-learning with extended Kalman filtering (EKF). Offline, a MAML-style meta-learning procedure optimizes adaptable basis functions and parameters; online, EKF enables efficient and safety-aware dynamic calibration of the onboard vehicle dynamics model. The framework further incorporates data-driven modeling and model predictive control (MPC). Real-world validation on a full-scale off-road vehicle demonstrates a 37% reduction in state prediction error, a 29% improvement in path tracking accuracy, and a 98.5% success rate in emergency obstacle avoidance. These results substantiate substantial enhancements in prediction fidelity, control performance, and system robustness under complex, unknown terrain conditions.

Technology Category

• Intelligent Robots: State Estimation
• Search and Optimization: Metareasoning and Metaheuristics
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

High-speed off-road autonomous driving presents unique challenges due to complex, evolving terrain characteristics and the difficulty of accurately modeling terrain-vehicle interactions. While dynamics models used in model-based control can be learned from real-world data, they often struggle to generalize to unseen terrain, making real-time adaptation essential. We propose a novel framework that combines a Kalman filter-based online adaptation scheme with meta-learned parameters to address these challenges. Offline meta-learning optimizes the basis functions along which adaptation occurs, as well as the adaptation parameters, while online adaptation dynamically adjusts the onboard dynamics model in real time for model-based control. We validate our approach through extensive experiments, including real-world testing on a full-scale autonomous off-road vehicle, demonstrating that our method outperforms baseline approaches in prediction accuracy, performance, and safety metrics, particularly in safety-critical scenarios. Our results underscore the effectiveness of meta-learned dynamics model adaptation, advancing the development of reliable autonomous systems capable of navigating diverse and unseen environments. Video is available at: https://youtu.be/cCKHHrDRQEA

Problem

• Adapting dynamics models for off-road autonomous driving
• Improving terrain-vehicle interaction prediction accuracy
• Enhancing safety in unseen off-road environments

Innovation

• Meta-learned parameters for dynamics model adaptation
• Kalman filter-based online adaptation scheme
• Real-time adjustment of onboard dynamics model