Learning Lane Graph Representations for Motion Forecasting

TL;DR

This paper introduces a novel lane graph-based motion forecasting model that leverages structured map representations and actor-map interactions, significantly improving prediction accuracy on the Argoverse benchmark.

Contribution

It presents LaneGCN, a graph convolutional network that explicitly models lane topology and actor-map interactions for better motion forecasting.

Findings

Outperforms state-of-the-art on Argoverse benchmark

Effectively captures complex lane topology and actor interactions

Produces accurate and realistic multi-modal trajectories

Abstract

We propose a motion forecasting model that exploits a novel structured map representation as well as actor-map interactions. Instead of encoding vectorized maps as raster images, we construct a lane graph from raw map data to explicitly preserve the map structure. To capture the complex topology and long range dependencies of the lane graph, we propose LaneGCN which extends graph convolutions with multiple adjacency matrices and along-lane dilation. To capture the complex interactions between actors and maps, we exploit a fusion network consisting of four types of interactions, actor-to-lane, lane-to-lane, lane-to-actor and actor-to-actor. Powered by LaneGCN and actor-map interactions, our model is able to predict accurate and realistic multi-modal trajectories. Our approach significantly outperforms the state-of-the-art on the large scale Argoverse motion forecasting benchmark.