Logo image
Back
ReasonDrive: Efficient Visual Question Answering for Autonomous Vehicles with Reasoning-Enhanced Small Vision-Language Models
Conference proceeding

ReasonDrive: Efficient Visual Question Answering for Autonomous Vehicles with Reasoning-Enhanced Small Vision-Language Models

Amirhosein Chahe and Lifeng Zhou
IEEE Computer Society Conference on Computer Vision and Pattern Recognition workshops, pp 3870-3879
11 Jun 2025
DOI: https://doi.org/10.1109/CVPRW67362.2025.00372

Abstract

Accuracy Adaptation models Autonomous Driving Chain-of-Thought Cognition Computational modeling Measurement Model Distillation Roads Robustness Vision-Language Models Visualization Autonomous Vehicles Safety
Vision-language models (VLMs) show promise for autonomous driving but often lack transparent reasoning capabilities that are critical for safety. We investigate whether explicitly modeling reasoning during fine-tuning enhances VLM performance on driving decision tasks. Using GPT-4o, we generate structured reasoning chains for driving scenarios from the DriveLM benchmark with category-specific prompting strategies. We compare reasoning-based fine-tuning, answer-only fine-tuning, and baseline instructiontuned models across multiple small VLM families (Llama 3.2, Llava 1.5, and Qwen 2.5VL). Our results demonstrate that reasoning-based fine-tuning consistently outperforms alternatives, with Llama3.2-11B-reason achieving the highest performance. Models fine-tuned with reasoning show substantial improvements in accuracy and text generation quality, suggesting explicit reasoning enhances internal representations for driving decisions. These findings highlight the importance of transparent decision processes in safety-critical domains and offer a promising direction for developing more interpretable autonomous driving systems. The code and dataset is available at https://github.com/Zhourobotics/ReasonDrive.

Metrics

1 Record Views

Details

Logo image