With the rapid development of ridesharing and connected and automated vehicles (CAVs), there are not only new intelligent transportation system (ITS) challenges with the emerging types of mobility, but there is also a huge opportunity to accelerate the connectivity and informatization of the transportation system, considering all new forms of data we get. For example, the origin-destination logs for ride-hailing trips can help us understand the human activity patterns and thus help to make demand predictions and create and improve the fleet dispatching system. While developing CAV applications, a large variety of data would be generated because of the sensors equipped for CAVs like LiDAR, GPS, and cameras. Data mining and machine learning technologies are the key tools help us process and discover knowledge from these new types of data. This research project is aimed to develop methods and applications for shared and autonomous mobility with machine learning and data mining knowledge and tools. The goal is to achieve a safer, faster, and more eco-friendly transportation system.
There are three main parts of this research:
- Lane-level road feature mapping by data mining crowdsourced GPS trajectories.
- Data-driven ride-hailing demand prediction, dispatching operation, system efficiency evaluation.
- Deep learning based real-time computer vision system for lane change behavior detection