ENAC Seminar Series by Prof. R. Pedarsani
Event details
| Date | 06.02.2020 |
| Hour | 09:45 › 10:45 |
| Speaker | Prof. Ramtin Pedarsani |
| Location | |
| Category | Conferences - Seminars |
09:45 – 10:45 – Prof. Ramtin Pedarsani
Assistant Professor, University of California, Santa Barbara, USA
Traffic Networks with Mixed Autonomy: Analysis, Learning, and Control
Traffic congestion has large economic and social costs. The introduction of autonomous vehicles can potentially reduce this congestion by increasing road capacity via vehicle platooning and by creating an avenue for influencing people’s choice of routes. In this talk, we consider traffic networks with mixed autonomy where a fraction of vehicles are human-driven and the rest are autonomous. We formalize a model of vehicle flow in mixed autonomy based on the fundamental diagram of traffic. We consider a network of parallel roads, characterize user equilibria, and provide a polynomial-time algorithm that computes optimal equilibria. Incorporating autonomous ride-hailing services in our model, we next develop an active preference-based learning algorithm to learn how people value time and money in choosing their preferred transportation option. This enables learning a model for people’s routing choices in a data-efficient manner. We then formulate a planning optimization that chooses service prices to maximize a social objective. We demonstrate the benefit of the proposed scheme by comparing the results to theoretical benchmarks. Finally, we study a dynamic routing game, in which the route choices of autonomous cars can be controlled and the human drivers react selfishly and dynamically to autonomous cars’ actions. As the problem dimension is prohibitively large, we use deep reinforcement learning to learn a policy for controlling the autonomous vehicles. This policy influences human drivers to route themselves in such a way that minimizes congestion on the network.
Assistant Professor, University of California, Santa Barbara, USA
Traffic Networks with Mixed Autonomy: Analysis, Learning, and Control
Traffic congestion has large economic and social costs. The introduction of autonomous vehicles can potentially reduce this congestion by increasing road capacity via vehicle platooning and by creating an avenue for influencing people’s choice of routes. In this talk, we consider traffic networks with mixed autonomy where a fraction of vehicles are human-driven and the rest are autonomous. We formalize a model of vehicle flow in mixed autonomy based on the fundamental diagram of traffic. We consider a network of parallel roads, characterize user equilibria, and provide a polynomial-time algorithm that computes optimal equilibria. Incorporating autonomous ride-hailing services in our model, we next develop an active preference-based learning algorithm to learn how people value time and money in choosing their preferred transportation option. This enables learning a model for people’s routing choices in a data-efficient manner. We then formulate a planning optimization that chooses service prices to maximize a social objective. We demonstrate the benefit of the proposed scheme by comparing the results to theoretical benchmarks. Finally, we study a dynamic routing game, in which the route choices of autonomous cars can be controlled and the human drivers react selfishly and dynamically to autonomous cars’ actions. As the problem dimension is prohibitively large, we use deep reinforcement learning to learn a policy for controlling the autonomous vehicles. This policy influences human drivers to route themselves in such a way that minimizes congestion on the network.
Practical information
- General public
- Free
Organizer
- ENAC
Contact
- Cristina Perez