Researchers deployed 100 RL-controlled cars to smooth congestion and reduce fuel consumption on highways. These AVs learned to mitigate stop-and-go traffic waves through data-driven simulations and real-world tests, achieving significant energy savings. The field test demonstrated the scalability and potential of this approach

10m read timeFrom bair.berkeley.edu
Post cover image
Table of contents
The challenges of phantom jamsReinforcement learning for wave-smoothing AVs100 AV field test: deploying RL at scaleFinal thoughts

Sort: