🤖 AI Summary
To address the lack of dynamical modeling and geometric equivariance in multi-agent motion prediction for autonomous driving, this paper proposes the first equivariant interaction modeling framework tailored for vehicle trajectory prediction. Methodologically, we introduce EqMotion—an equivariant particle-based model—into this domain for the first time, integrating equivariant graph neural networks with multimodal conditional variational inference to strictly enforce motion equivariance and interaction invariance under Euclidean transformations. Our contributions are threefold: (1) a unified architecture that jointly achieves geometric awareness and interaction-aware modeling; (2) state-of-the-art accuracy on major benchmarks (nuScenes, Argoverse) with only 1.2M parameters; and (3) significantly improved training efficiency—converging in under two hours on a single GPU.
📝 Abstract
Forecasting vehicular motions in autonomous driving requires a deep understanding of agent interactions and the preservation of motion equivariance under Euclidean geometric transformations. Traditional models often lack the sophistication needed to handle the intricate dynamics inherent to autonomous vehicles and the interaction relationships among agents in the scene. As a result, these models have a lower model capacity, which then leads to higher prediction errors and lower training efficiency. In our research, we employ EqMotion, a leading equivariant particle, and human prediction model that also accounts for invariant agent interactions, for the task of multi-agent vehicle motion forecasting. In addition, we use a multi-modal prediction mechanism to account for multiple possible future paths in a probabilistic manner. By leveraging EqMotion, our model achieves state-of-the-art (SOTA) performance with fewer parameters (1.2 million) and a significantly reduced training time (less than 2 hours).