From Goals, Waypoints & Paths To Long Term Human Trajectory Forecasting

TL;DR

This paper introduces a novel approach to long-term human trajectory forecasting by modeling epistemic and aleatoric uncertainties separately, achieving state-of-the-art results on both short and long prediction horizons.

Contribution

It proposes a new factorization of uncertainty into epistemic and aleatoric components and a novel long-term forecasting setting with horizons up to a minute.

Findings

Y-net outperforms previous models on short horizon datasets.

Y-net achieves significant improvements on the new long horizon datasets.

The approach effectively captures diverse future trajectories.

Abstract

Human trajectory forecasting is an inherently multi-modal problem. Uncertainty in future trajectories stems from two sources: (a) sources that are known to the agent but unknown to the model, such as long term goals and (b)sources that are unknown to both the agent & the model, such as intent of other agents & irreducible randomness indecisions. We propose to factorize this uncertainty into its epistemic & aleatoric sources. We model the epistemic un-certainty through multimodality in long term goals and the aleatoric uncertainty through multimodality in waypoints& paths. To exemplify this dichotomy, we also propose a novel long term trajectory forecasting setting, with prediction horizons upto a minute, an order of magnitude longer than prior works. Finally, we presentY-net, a scene com-pliant trajectory forecasting network that exploits the pro-posed epistemic & aleatoric structure for diverse trajectory predictions across long prediction horizons.Y-net significantly improves previous state-of-the-art performance on both (a) The well studied short prediction horizon settings on the Stanford Drone & ETH/UCY datasets and (b) The proposed long prediction horizon setting on the re-purposed Stanford Drone & Intersection Drone datasets.