ForceFormer: Exploring Social Force and Transformer for Pedestrian Trajectory Prediction

Abstract

Predicting trajectories of pedestrians based on goal information in highly interactive scenes is a crucial step toward Intelligent Transportation Systems and Autonomous Driving. The challenges of this task come from two key sources:(1) complex social interactions in high pedestrian density scenarios and ( 2) limited utilization of goal information to effectively associate with past motion information. To address these difficulties, we integrate social forces into a Transformerbased stochastic generative model backbone and propose a new goal-based trajectory predictor called ForceFormer. Differentiating from most prior works that simply use the destination position as an input feature, we leverage the driving force from the destination to efficiently simulate the guidance of a target on a pedestrian. Additionally, repulsive forces are used as another input feature to describe the avoidance action among neighboring pedestrians. Extensive experiments show that our proposed method achieves on-par performance measured by distance errors with the state-of-the-art models but evidently decreases collisions, especially in dense pedestrian scenarios on widely used pedestrian datasets.

A Review of Trajectory Prediction Methods for the Vulnerable Road User

“…For the pedestrian datasets and Argoverse 1, anchor-conditioned models also showed outstanding performance metrics [21,23,25,26,28,29,31]. Again, their setup allows them to model multimodality very well.…”

“…Anchor-conditioned models use a set of anchor points to condition the prediction of an agent's trajectory [21,[26][27][28][29][30][31]. These anchor points can be, for example, the respective agent's estimated final goal position, which is used to condition the final trajectory estimation or even a trajectory proposal.…”

“…Zhang et al [21] presented a goal-conditioned extension of AgentFormer [49] with ForceFormer. The authors extended the original architecture with a goal estimation module and a social force module.…”