AI learns to route traffic faster by adapting decision-making on the fly

Researchers have developed a reinforcement learning system that optimizes how vehicles share route information in real time, automatically adjusting how many candidate paths to evaluate based on network conditions. The advance could cut delays in congested areas while reducing the computational burden on vehicles—a key concern as autonomous fleets scale up.

Originaltitel: Adaptively tuning candidates forwarding set sizes via extended Q-learning in opportunistic vehicular routing schemes