Reinforcement Learning for Mixed-Autonomy Traffic
Cathy Wu, Eugene Vinitsky

Citation
Cathy Wu, Eugene Vinitsky. "Reinforcement Learning for Mixed-Autonomy Traffic". Talk or presentation, 23, August, 2017.

Abstract
We study the mixed-autonomy traffic control problem using the deep reinforcement learning framework. This model-free learning method turns out to select policies and behaviors previously discovered by model-driven approaches, such as stabilization, platooning, and vehicle bunching, known to improve ring road and intersection efficiency. Remarkably, by effectively leveraging the structure of the human driving behavior, the learned policies additionally surpass the performance of state-of-the-art controllers designed for automated vehicles. Finally, we provide scaling of our results to large action spaces via the introduction of state-action equivalence classes.

Electronic downloads


(No downloads are available for this publication.)
Citation formats  
  • HTML
    Cathy Wu, Eugene Vinitsky. <a
    href="http://www.cps-forces.org/pubs/265.html"
    ><i>Reinforcement Learning for Mixed-Autonomy
    Traffic</i></a>, Talk or presentation,  23,
    August, 2017.
  • Plain text
    Cathy Wu, Eugene Vinitsky. "Reinforcement Learning for
    Mixed-Autonomy Traffic". Talk or presentation,  23,
    August, 2017.
  • BibTeX
    @presentation{WuVinitsky17_ReinforcementLearningForMixedAutonomyTraffic,
        author = {Cathy Wu and Eugene Vinitsky},
        title = {Reinforcement Learning for Mixed-Autonomy Traffic},
        day = {23},
        month = {August},
        year = {2017},
        abstract = {We study the mixed-autonomy traffic control
                  problem using the deep reinforcement learning
                  framework. This model-free learning method turns
                  out to select policies and behaviors previously
                  discovered by model-driven approaches, such as
                  stabilization, platooning, and vehicle bunching,
                  known to improve ring road and intersection
                  efficiency. Remarkably, by effectively leveraging
                  the structure of the human driving behavior, the
                  learned policies additionally surpass the
                  performance of state-of-the-art controllers
                  designed for automated vehicles. Finally, we
                  provide scaling of our results to large action
                  spaces via the introduction of state-action
                  equivalence classes.},
        URL = {http://cps-forces.org/pubs/265.html}
    }
    

Posted by Carolyn Winter on 24 Aug 2017.
Groups: forces
For additional information, see the Publications FAQ or contact webmaster at cps-forces org.

Notice: This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright.