This research leverages vision-language models to improve heavy-duty truck classification from point cloud data, addressing critical safety challenges in autonomous driving without extensive manual annotations.

• Combines LiDAR point cloud data with advanced language model capabilities
• Reduces dependency on labor-intensive manual annotation processes
• Enhances classification accuracy for vehicle safety systems
• Provides framework for more efficient cooperative autonomous driving

Security Impact: By improving heavy-duty truck classification, this approach directly enhances road safety through better vehicle identification and tracking, enabling autonomous systems to make more informed decisions when interacting with larger vehicles that pose greater safety risks.

Investigating Vision-Language Model for Point Cloud-based Vehicle Classification