At 16:00pm on July 8th, the forth phase of the "Engineering Management Forum" series of lectures of School of Economics and Management of Beihang University was successfully held on the Tencent conference platform. In this lecture, Professor Jiancheng Long, a doctoral supervisor a professor atthe School of Automobile and Transportation Engineering of Hefei University of Technology,gave us a detailed description of a time-dependent shared autonomous vehicle system design problem
Professor Jiancheng Long is theYoung Changjiang Scholars Distinguished Professor of the Ministry of Education.He has presided over a number of scientific research projects, such as the National Science Fund for Distinguished Young Scholars (2019) and the National Science Fund for Excellent Young Scholars (2015). He is currently a director of Society of Management Science and Engineering of China, a director of Systems Engineering Society of China, vice president of the Nonlinear Science Society of Anhui Province and an editorial board member of the SCI/SSCI Journal“International Journal of Transportation”and“Control and Decision”.
Professor Jiancheng Long proposes a time-dependent SAV system design problem by jointly optimizing fleet size, parking infrastructure deployment, and daily operation of the system for infrastructure planning in the long run. The dynamic system optimum (DSO) principle in terms of total daily system cost (TDSC) is adopted to formulate the daily operation of the SAV system, i.e., users’ departure time choices and SAVs’ route choices. By incorporating the link transmission model (LTM) as the traffic flow model, the daily operation problem (DOP) of the SAV system is formulated as a linear programming (LP) problem. Further, the time-dependent SAV system design problem is formulated as a mixed integer linear programming (MILP) problem. The LP relaxation of the proposed MILP problem could provide a tight lower bound, and a diving heuristic algorithm is developed to solve the proposed MILP problem. Finally, numerical examples are designed to illustrate the properties of the model and the efficiency of the proposed solution algorithm.
The results show that the proposed diving heuristic algorithm can achieve nearly optimal solutions for the proposed MILP problem in a timely fashion. The developed methodology is robust in terms of changing model parameters, such as the fixed cost of constructing parking lots, the maximum number of parking spaces, and the purchase price of one SAV. Besides, the results show that the developed SAV system is capable of balancing between parking lot construction cost and access to pick up travelers by arranging some distant parking locations from CBDs or residential areas with high land costs. Compared with the UAV system, the proposed SAV system can greatly save fleet size and parking spaces with a limited increase in TSTC, especially when the travel demand level is high. Compared with the SAV system without long-run planning, the proposed SAV system with long-run planning has much lower DTSC, especially when the number of planning years grows up.
Finally, Professor Jiancheng Long suggested some future research.It should be noted that the scale of the problem is still challenging under the current framework because solving the LP relaxation of the proposed MILP problem is time-consuming for large-scale realistic networks. To improve computational efficiency, some decomposition methods, such as Dantzig-Wolfe decomposition, can be explored to solve the LP sub-problems more efficiently. Besides, this paper assumes homogeneous travelers, which can be extended into heterogeneous cases. For example, travelers are heterogeneous in terms of the value of travel time, early or late arrival penalties, time windows. The incorporation of these factors makes the model step closer to the real world. This study also motivates some further research directions for SAV systems, such as considering ride-sharing, demand uncertainty, and multiclass vehicles in the proposed model.
Professor Jiancheng Long’s report with clear logic and simple explanations benefited everyone. During the meeting, teachers and students had in-depth discussions with Professor Long about some puzzles in this research. The report was successfully concluded in a warm atmosphere.
