A 2B VLM fine-tuned in two stages on a physics-encoded bird's-eye view image outperforms a zero-shot 8B VLM on eight driving-reasoning question types and raises emergency-braking recall from 10.8% to 47.8%.
VLA-MP: A vision-language-action framework for multimodal perception and physics-constrained action generation in autonomous driving,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MoRAL: Sensor-Grounded BEV Reasoning for Compact VLMs toward Edge-Oriented Autonomous Driving
A 2B VLM fine-tuned in two stages on a physics-encoded bird's-eye view image outperforms a zero-shot 8B VLM on eight driving-reasoning question types and raises emergency-braking recall from 10.8% to 47.8%.