Deep Learning-Based Optimization of Eco-Driving Strategies with Connected and Autonomous Electric Vehicles on Transportation Networks

Connected and Autonomous Electric Vehicles (CAEVs) allow the design of more advanced driving strategies, such as eco-driving strategies, towards even lower energy consumption when passing intersections. This project will create a deep learning-based optimization system on eco-driving strategies for traffic operations over transportation networks with CAEVs under complicated dynamic traffic conditions. The research will start with collecting field operational data on CAEV in transportation networks (isolated intersections, arterial streets, and road networks). Then, deep learning-based algorithms will be developed to optimize the eco-driving strategies for CAEVs on various transportation networks, while the field test data will validate the optimization model. Finally, the driving simulator tests will be conducted to measure the optimization models and eco-driving strategies. About twenty subjects will be recruited for the simulator tests over different simulation scenes: with and without the optimized eco-driving strategies. The deep learning-based optimization of eco-driving strategy is expected to significantly reduce the energy consumption of CAEVs at isolated intersections, along arterial streets, and on road networks. This project will help both transportation and environmental agencies at all levels, and car manufacturers, to understand the design, operation, and impacts of optimal eco-driving strategies. The project will provide urgent science and test-based input to inform policy and practice development.

Research Area