Accurate modeling and prediction of driving behavior are crucial for enabling autonomous vehicles to safely navigate complex, interactive traffic environments. While recent continual learning approaches for interactive trajectory prediction aim to learn efficiently from streaming data, they often fail to fully retain previously learned cases when acquiring new knowledge, a phenomenon we term case-level forgetting. This limitation poses significant risks in safety-critical autonomous driving applications. This paper identifies, analyzes, and addresses case-level forgetting in continual learning for trajectory prediction. We propose the Dynamically Expandable Interactive Trajectory Predictor (DEITP), a novel framework that preserves previously learned knowledge through a dynamic model expansion mechanism. The mechanism regulates expansion timing by assessing model similarity, thereby controlling model growth while preventing catastrophic forgetting. Furthermore, to operate in realistic task-free settings where task identity is unavailable at test time, we introduce a task identification strategy based on a familiarity autoencoder that selects the most appropriate expert for prediction. Extensive experiments on real-world datasets demonstrate that DEITP substantially mitigates forgetting and achieves zero-forgetting performance when task identities are known.
Li et al. (2026) studied this question.