Agent-based simulator for large-scale shared mobility systems
We developed an agent-based simulator that incorporates real-world transportation road networks and the detailed operation dynamics of individual shared mobility vehicles. The tool can be used as the environment for developing reinforcement learning algorithms or as an analytical tool for understanding the efficiency, dynamics, and resilience of autonomous/electric shared mobility systems.
The code is open to the public: DROP_Simulator
The code is open to the public: DROP_Simulator
Algorithm and data for disease-risk minimizing vehicle routing problem
We developed a branch-price-cut algorithm to solve the dial-a-ride problem with minimum disease transmission risk. The algorithm can minimize total operation cost while significantly reducing contagion risk among passengers. We tested the algorithm on classic DARP cases and found that over 90% reduction in risk can be achieved with a ~13% increase in cost. We additionally generated instances from real-world paratransit operation services for benchmark purposes.
The code and data are open to the public at: RDARP algo & RDARP data repo
The code and data are open to the public at: RDARP algo & RDARP data repo