Multimodal trajectory forecasting based on discrete heat map

TL;DR

This paper introduces a multimodal trajectory forecasting method using heat maps, leveraging lane maps and historical data to predict multiple probable future paths in traffic scenes.

Contribution

The work presents a novel heat map-based approach for multimodal trajectory prediction that effectively models probabilistic future trajectories in traffic environments.

Findings

Achieved accurate probabilistic trajectory predictions in the Argoverse competition.

Predicted multiple plausible future trajectories for each target.

Demonstrated effectiveness of heat map representation in trajectory forecasting.

Abstract

In Argoverse motion forecasting competition, the task is to predict the probabilistic future trajectory distribution for the interested targets in the traffic scene. We use vectorized lane map and 2 s targets' history trajectories as input. Then the model outputs 6 forecasted trajectories with probability for each target.