BANet: Motion Forecasting with Boundary Aware Network

TL;DR

BANet enhances motion forecasting by incorporating comprehensive vector map elements, including lane boundaries, to better encode traffic rules and improve prediction accuracy, achieving top performance in the 2022 Argoverse2 challenge.

Contribution

The paper introduces BANet, a novel boundary-aware network that encodes multiple vector map elements beyond lane centerlines for improved motion forecasting.

Findings

Achieved state-of-the-art results on Argoverse2 challenge

Incorporating lane boundary information improves prediction accuracy

Outperforms previous lane-centric models

Abstract

We propose a motion forecasting model called BANet, which means Boundary-Aware Network, and it is a variant of LaneGCN. We believe that it is not enough to use only the lane centerline as input to obtain the embedding features of the vector map nodes. The lane centerline can only provide the topology of the lanes, and other elements of the vector map also contain rich information. For example, the lane boundary can provide traffic rule constraint information such as whether it is possible to change lanes which is very important. Therefore, we achieved better performance by encoding more vector map elements in the motion forecasting model.We report our results on the 2022 Argoverse2 Motion Forecasting challenge and rank 1st on the test leaderboard.