LatentFormer: Multi-Agent Transformer-Based Interaction Modeling and Trajectory Prediction
This addresses the problem of predicting vehicle trajectories in autonomous driving, offering a novel method for interaction modeling that could enhance safety and planning, though it appears incremental as it builds on existing transformer-based approaches.
The paper tackled multi-agent trajectory prediction for autonomous driving by proposing LatentFormer, a transformer-based model that models interactions using future states and multi-resolution map encoding, achieving state-of-the-art performance with up to 40% improvement in trajectory metrics on the nuScenes dataset.
Multi-agent trajectory prediction is a fundamental problem in autonomous driving. The key challenges in prediction are accurately anticipating the behavior of surrounding agents and understanding the scene context. To address these problems, we propose LatentFormer, a transformer-based model for predicting future vehicle trajectories. The proposed method leverages a novel technique for modeling interactions among dynamic objects in the scene. Contrary to many existing approaches which model cross-agent interactions during the observation time, our method additionally exploits the future states of the agents. This is accomplished using a hierarchical attention mechanism where the evolving states of the agents autoregressively control the contributions of past trajectories and scene encodings in the final prediction. Furthermore, we propose a multi-resolution map encoding scheme that relies on a vision transformer module to effectively capture both local and global scene context to guide the generation of more admissible future trajectories. We evaluate the proposed method on the nuScenes benchmark dataset and show that our approach achieves state-of-the-art performance and improves upon trajectory metrics by up to 40%. We further investigate the contributions of various components of the proposed technique via extensive ablation studies.