Browse Teach Curate

Semantic Vehicular Network Transmission with Vision-Language Models and Deep Reinforcement Learning

An adaptive transmission framework for vehicular networks that uses a vision-language model to extract task-critical semantic content from vehicle camera images (reducing data volume over 90 percent) and a GAE-PPO deep reinforcement learning controller to select semantic symbol length, with the Weber-Fechner psychophysical law as the quality-of-experience metric balancing bandwidth against perceived quality.

B9 Text

JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2015 1 Embodied AI-Enhanced Vehicular

This paper proposes an adaptive transmission framework for vehicular networks that combines semantic information extraction with deep reinforcement learning. A vision-language model (LLAVA) extracts …