Interventional Behavior Prediction: Avoiding Overly Confident Anticipation in Interactive Prediction

TL;DR

This paper introduces interventional behavior prediction (IBP) to address overconfidence in conditional behavior prediction models by treating planned trajectories as interventions, and proposes a Shapley-value-based metric for evaluation.

Contribution

It proposes the IBP framework that models behavior under interventions and introduces a Shapley-value-based metric to evaluate temporal independence in predictions.

Findings

The Shapley-value metric effectively detects violations of temporal independence.

IBP provides a more realistic prediction framework by modeling interventions.

The approach enhances safety and reliability in interactive autonomous systems.

Abstract

Conditional behavior prediction (CBP) builds up the foundation for a coherent interactive prediction and planning framework that can enable more efficient and less conservative maneuvers in interactive scenarios. In CBP task, we train a prediction model approximating the posterior distribution of target agents' future trajectories conditioned on the future trajectory of an assigned ego agent. However, we argue that CBP may provide overly confident anticipation on how the autonomous agent may influence the target agents' behavior. Consequently, it is risky for the planner to query a CBP model. Instead, we should treat the planned trajectory as an intervention and let the model learn the trajectory distribution under intervention. We refer to it as the interventional behavior prediction (IBP) task. Moreover, to properly evaluate an IBP model with offline datasets, we propose a Shapley-value-based metric to verify if the prediction model satisfies the inherent temporal independence of an interventional distribution. We show that the proposed metric can effectively identify a CBP model violating the temporal independence, which plays an important role when establishing IBP benchmarks.