Domain Knowledge Driven Pseudo Labels for Interpretable Goal-Conditioned Interactive Trajectory Prediction
This work addresses motion forecasting for autonomous vehicles in interactive scenarios, representing an incremental improvement by enhancing interpretability and performance in a specific domain.
The paper tackles the problem of joint trajectory prediction in interactive autonomous driving scenarios by proposing a goal-conditioned CVAE model that uses domain knowledge-driven pseudo labels to avoid posterior collapse and induce an interpretable latent space, validated on the Waymo Open Motion Dataset with quantitative and qualitative evaluations.
Motion forecasting in highly interactive scenarios is a challenging problem in autonomous driving. In such scenarios, we need to accurately predict the joint behavior of interacting agents to ensure the safe and efficient navigation of autonomous vehicles. Recently, goal-conditioned methods have gained increasing attention due to their advantage in performance and their ability to capture the multimodality in trajectory distribution. In this work, we study the joint trajectory prediction problem with the goal-conditioned framework. In particular, we introduce a conditional-variational-autoencoder-based (CVAE) model to explicitly encode different interaction modes into the latent space. However, we discover that the vanilla model suffers from posterior collapse and cannot induce an informative latent space as desired. To address these issues, we propose a novel approach to avoid KL vanishing and induce an interpretable interactive latent space with pseudo labels. The proposed pseudo labels allow us to incorporate domain knowledge on interaction in a flexible manner. We motivate the proposed method using an illustrative toy example. In addition, we validate our framework on the Waymo Open Motion Dataset with both quantitative and qualitative evaluations.