Commuter ride-sharing using topology-based vehicle trajectory clustering: Methodology, application and impact evaluation

You are here: Homepage > Search

Commuter ride-sharing using topology-based vehicle trajectory clustering: Methodology, application and impact evaluation

Hong, Zihan / Chen, Ying / Mahmassani, Hani S. / Xu, Shuang

Highlights•Algorithm for clustering trajectories provides basis for ride-matching commuters.•Recommendations based on matched trajectories more likely to be followed.•Algorithm-recommended shared rides evaluated in Chicago test network.•Evaluation shows performance improvement in delay and sustainability.

AbstractThis paper illustrates a ride matching method for commuting trips based on clustering trajectories, and a modeling and simulation framework with ride-sharing behaviors to illustrate its potential impact. It proposes data mining solutions to reduce traffic demand and encourage more environment-friendly behaviors. The main contribution is a new data-driven ride-matching method, which tracks personal preferences of road choices and travel patterns to identify potential ride-sharing routes for carpool commuters. Compared with prevalent carpooling algorithms, which allow users to enter departure and destination information for on-demand trips, the proposed method focuses more on regular commuting trips. The potential effectiveness of the approach is evaluated using a traffic simulation-assignment framework with ride-sharing participation using the routes suggested by our algorithm. Two types of ride-sharing participation scenarios, with and without carpooling information, are considered. A case study with the Chicago tested is conducted to demonstrate the proposed framework’s ability to support better decision-making for carpool commuters. The results indicate that with ride-matching recommendations using shared vehicle trajectory data, carpool programs for commuters contribute to a less congested traffic state and environment-friendly travel patterns.