Identifying Driver Interactions via Conditional Behavior Prediction

TL;DR

This paper introduces end-to-end models for conditional behavior prediction in interactive driving scenarios, enabling better modeling of agent reactions and scoring of scenario interactivity for training and evaluation.

Contribution

It presents a novel CBP model that predicts other agents' trajectories conditioned on ego-agent plans and introduces an interactivity score for scenario analysis.

Findings

The CBP model accurately predicts agent trajectories conditioned on ego plans.

The interactivity score effectively identifies interesting interactive scenarios.

The score improves agent prioritization under computational constraints.

Abstract

Interactive driving scenarios, such as lane changes, merges and unprotected turns, are some of the most challenging situations for autonomous driving. Planning in interactive scenarios requires accurately modeling the reactions of other agents to different future actions of the ego agent. We develop end-to-end models for conditional behavior prediction (CBP) that take as an input a query future trajectory for an ego-agent, and predict distributions over future trajectories for other agents conditioned on the query. Leveraging such a model, we develop a general-purpose agent interactivity score derived from probabilistic first principles. The interactivity score allows us to find interesting interactive scenarios for training and evaluating behavior prediction models. We further demonstrate that the proposed score is effective for agent prioritization under computational budget constraints.