HiLM-D introduces a multi-scale high-resolution approach that enables autonomous vehicles to identify risks and provide driving suggestions using natural language.

• Implements ROLISP: Risk Object Localization and Intention and Suggestion Prediction
• Addresses the perception gap in autonomous driving systems
• Leverages multi-scale high-resolution details for improved environmental understanding
• Enhances safety and interpretability of autonomous driving decisions

This engineering breakthrough matters because it brings human-like reasoning to vehicle systems, potentially reducing accidents through better risk assessment and more intuitive driver-vehicle communication.

HiLM-D: Enhancing MLLMs with Multi-Scale High-Resolution Details for Autonomous Driving