MetaDAT: Generalizable Trajectory Prediction Via Meta Pre-Training and Data-Adaptive Test-Time Updating
Yuning Wang, Pu Zhang, Yuan He, Ke Wang, Jianru Xue
AI summary
Problem
Offline-trained trajectory predictors degrade under test-time distribution shifts because pre-training objectives misalign with online adaptation goals, and fixed updating rules cannot adapt to unknown test data characteristics.
Approach
MetaDAT uses meta-learning to pre-train a flexible model initialization optimized for online adaptation, then dynamically adjusts learning rates and focuses updates on hard samples during test time.
Key results
- Surpasses state-of-the-art TTT methods by ~12.7% on mADE6 across cross-dataset shifts
- Maintains high accuracy under suboptimal learning rates and strict FPS constraints
- Effectively generalizes across nuScenes, Lyft, and Waymo for short- and long-term prediction
- Validates that meta pre-training and dynamic learning rate optimization jointly drive adaptation gains
Why it matters
Provides a robust, tunable-free adaptation framework for autonomous driving systems facing real-world environmental changes.
Abstract
Existing trajectory prediction methods exhibit significant performance degradation under distribution shifts during test time. Although test-time training techniques have been explored to enable adaptation, current approaches rely on an offline pre-trained predictor that lacks online learning flexibility. Moreover, they depend on fixed online model updating rules that do not accommodate the specific characteristics of test data. To address these limitations, we first propose a meta-learning framework to directly optimize the predictor for fast and accurate online adaptation, which performs bi- level optimization on the performance of simulated test-time adaptation tasks during pre-training. Furthermore, at test time, we introduce a data-adaptive model updating mechanism that dynamically adjusts the predefined learning rates and updating frequencies based on online partial derivatives and hard sample selection. This mechanism enables the online learning rate to suit the test data, and focuses on informative hard samples to enhance efficiency. Experiments are conducted on various challenging cross-dataset distribution shift scenarios, including nuScenes, Lyft, and Waymo. Results demonstrate that our method achieves superior adaptation accuracy, surpassing state-of-the-art test- time training methods for trajectory prediction. Additionally, our method excels under suboptimal learning rates and high FPS demands, showcasing its robustness and practicality.