Search papers, labs, and topics across Lattice.
This paper introduces DeLL, a Deconfounded Lifelong Learning framework for end-to-end autonomous driving that uses a Dirichlet process mixture model (DPMM) to create dynamic knowledge spaces representing driving behaviors and latent abilities. DeLL employs front-door adjustment to deconfound spurious correlations, improving causal representation learning and mitigating catastrophic forgetting. Experiments in CARLA, using newly proposed evaluation protocols, demonstrate DeLL's improved adaptability and driving performance in lifelong learning scenarios.
Autonomous driving systems can now learn continuously without forgetting, thanks to a new method that disentangles true driving skills from spurious correlations caused by sensor noise and environmental changes.
End-to-End autonomous driving (E2E-AD) systems face challenges in lifelong learning, including catastrophic forgetting, difficulty in knowledge transfer across diverse scenarios, and spurious correlations between unobservable confounders and true driving intents. To address these issues, we propose DeLL, a Deconfounded Lifelong Learning framework that integrates a Dirichlet process mixture model (DPMM) with the front-door adjustment mechanism from causal inference. The DPMM is employed to construct two dynamic knowledge spaces: a trajectory knowledge space for clustering explicit driving behaviors and an implicit feature knowledge space for discovering latent driving abilities. Leveraging the non-parametric Bayesian nature of DPMM, our framework enables adaptive expansion and incremental updating of knowledge without predefining the number of clusters, thereby mitigating catastrophic forgetting. Meanwhile, the front-door adjustment mechanism utilizes the DPMM-derived knowledge as valid mediators to deconfound spurious correlations, such as those induced by sensor noise or environmental changes, and enhances the causal expressiveness of the learned representations. Additionally, we introduce an evolutionary trajectory decoder that enables non-autoregressive planning. To evaluate the lifelong learning performance of E2E-AD, we propose new evaluation protocols and metrics based on Bench2Drive. Extensive evaluations in the closed-loop CARLA simulator demonstrate that our framework significantly improves adaptability to new driving scenarios and overall driving performance, while effectively retaining previous acquired knowledge.
