Evaluation of Drivers' Interaction Ability at Social Scenarios: A Process-Based Framework

TL;DR

This paper addresses the challenge of evaluating drivers' interactive abilities in dynamic social driving contexts by proposing a three-stage, process-based framework: Risk Perception Modeling, Interactive Process Modeling, and Interactive Ability Scoring. It integrates motion-state estimation with risk field theory, plus game-theoretic rational-agent benchmarks (non-cooperative and cooperative) to model and assess interaction dynamics, and introduces an improved morphological similarity distance to score drivers relative to rational benchmarks. The approach is validated on unsignalized intersections using Chinese and US driving datasets, demonstrating the framework’s ability to distinguish conservative versus aggressive interactions and its adaptability across regional contexts. The work contributes a scalable, interpretable evaluation tool for driving risk, efficiency, and social interaction, with potential to guide autonomous vehicle behavior in human-driven environments and to support real-time assessment of driver interaction capabilities.

Abstract

Assessing drivers' interaction capabilities is crucial for understanding human driving behavior and enhancing the interactive abilities of autonomous vehicles. In scenarios involving strong interaction, existing metrics focused on interaction outcomes struggle to capture the evolutionary process of drivers' interactive behaviors, making it challenging for autonomous vehicles to dynamically assess and respond to other agents during interactions. To address this issue, we propose a framework for assessing drivers' interaction capabilities, oriented towards the interactive process itself, which includes three components: Interaction Risk Perception, Interaction Process Modeling, and Interaction Ability Scoring. We quantify interaction risks through motion state estimation and risk field theory, followed by introducing a dynamic action assessment benchmark based on a game-theoretical rational agent model, and designing a capability scoring metric based on morphological similarity distance. By calculating real-time differences between a driver's actions and the assessment benchmark, the driver's interaction capabilities are scored dynamically. We validated our framework at unsignalized intersections as a typical scenario. Validation analysis on driver behavior datasets from China and the USA shows that our framework effectively distinguishes and evaluates conservative and aggressive driving states during interactions, demonstrating good adaptability and effectiveness in various regional settings.

Figures (11)

• Figure 1: The framework of the interaction ability evaluation.
• Figure 2: The interaction between one turn-left vehicle and one straight vehicle at the unprotected turning intersection.
• Figure 3: The left turn events from two datasets: (a) the Xianxia-Jianhe Intersection; (b) the intersection from Waymo Left-Turn Motion Dataset.
• Figure 4: The fitness of the objective function with different hyperparameters during the optimization process.
• Figure 5: Drivers’ average ability score from two datasets.