What Do They See? Interpreting Complex Road Scenarios Through the Eyes of Vision-Language-Action Models for Safe and Trustworthy Autonomous Vehicle Learning
Kalpana Panda, Wesley Maia, Vinti Agarwal, Ross Greer
Why It Matters
What makes this one worth your time
This research addresses the critical need for interpretability in autonomous vehicle systems, enhancing safety and trust in AI-driven technologies.
A novel framework for understanding object influence in autonomous driving models.
Summary
The paper introduces a counterfactual ablation framework called CVAA to analyze the causal influence of individual objects in road scenarios on the planning behavior of autonomous vehicle models, demonstrating its application on the Alpamayo 1 trajectory predictor across 210 driving scenes.
Key contributions
- Development of the Counterfactual Vision Action Analysis (CVAA) framework.
- Creation of the Counter-nuScenes dataset for evaluating model responses to object removal.
- Insights into the causal relationships between detected objects and model planning behavior.
Notable insights
- The framework uses photorealistic generative inpainting to create counterfactual scenarios, allowing for a detailed causal analysis of object influence.
- Findings reveal that models may respond to irrelevant objects, questioning their interpretative alignment with human perception.
Possible limitations
- Not stated in the abstract.
Abstract
arXiv:2607.16938v1 Announce Type: cross Abstract: End-to-end autonomous driving models are now able to navigate complex road scenarios, mapping raw sensor observations directly to observed paths for open-loop evaluation and often effective driving in closed-loop evaluation. Yet the internal logic of these safety-critical systems remains largely opaque, due to the complexity of traffic scenes. We propose a counterfactual ablation framework called Counterfactual Vision Action Analysis (CVAA) that systematically removes individual detected objects from front-camera images using photorealistic generative inpainting to prepare counterfactual sets to evaluate the difference in the model's response. This isolates the causal effect of each object's presence on the model's planning behaviour. Applied to the Alpamayo 1 trajectory predictor across 210 nuScenes driving scenes, we create a dataset Counter -nuScenes, using which we see that vehicles and pedestrians within the model's 'path' dominate causal influence as expected, while traffic lights, as expected, exert disproportionate effect relative to their image footprint. However, we also find cases where the model responds strongly to objects a human driver would consider irrelevant. This brings forth a deeper question: does the model itself view the scene as a sum of individual objects influencing the outcome, or does it encode an entirely different set of internal features that do not correspond to human-legible scene elements? To further understand this, we compare intermediate representations of original and inpainted image pairs using mechanistic interpretability techniques and examine the effect of the removal through the various model layers. Together, these two stages offer a path from behavioral auditing to representational understanding, creating explainable driving systems and solidifying human-AI trust.