CARE Drive A Framework for Evaluating Reason-Responsiveness of Vision Language Models in Automated Driving

2026-02-17

AI Summary

Current evaluation methods for vision-language models in autonomous driving focus solely on output outcomes, making it difficult to assess whether the models’ decisions genuinely respond to human-interpretable reasons and thus risking misjudgment of their explanatory capabilities. This work proposes CARE Drive, the first model-agnostic framework for systematic evaluation: it employs a two-stage design that first stabilizes outputs through prompt calibration and then applies controlled contextual perturbations to quantify the model’s causal responsiveness to normative reasons such as safe following distance and social pressure. Experiments in cyclist-overtaking scenarios demonstrate that explicitly incorporating human-aligned reasons significantly improves alignment with expert driving behavior; however, the models exhibit uneven sensitivity across different reasons, revealing persistent limitations in their capacity to reason responsively.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Common-Sense Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Foundation models, including vision language models, are increasingly used in automated driving to interpret scenes, recommend actions, and generate natural language explanations. However, existing evaluation methods primarily assess outcome based performance, such as safety and trajectory accuracy, without determining whether model decisions reflect human relevant considerations. As a result, it remains unclear whether explanations produced by such models correspond to genuine reason responsive decision making or merely post hoc rationalizations. This limitation is especially significant in safety critical domains because it can create false confidence. To address this gap, we propose CARE Drive, Context Aware Reasons Evaluation for Driving, a model agnostic framework for evaluating reason responsiveness in vision language models applied to automated driving. CARE Drive compares baseline and reason augmented model decisions under controlled contextual variation to assess whether human reasons causally influence decision behavior. The framework employs a two stage evaluation process. Prompt calibration ensures stable outputs. Systematic contextual perturbation then measures decision sensitivity to human reasons such as safety margins, social pressure, and efficiency constraints. We demonstrate CARE Drive in a cyclist overtaking scenario involving competing normative considerations. Results show that explicit human reasons significantly influence model decisions, improving alignment with expert recommended behavior. However, responsiveness varies across contextual factors, indicating uneven sensitivity to different types of reasons. These findings provide empirical evidence that reason responsiveness in foundation models can be systematically evaluated without modifying model parameters.

Problem

• reason-responsiveness
• vision language models
• automated driving
• model evaluation
• human-aligned reasoning

Innovation

• reason-responsiveness
• vision language models
• automated driving
• contextual perturbation
• model-agnostic evaluation