Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SC31

in real time without knowing when and where future requests will occur. We show dynamic strategies to set time windows as a function of arrival time variance perform significantly better than static policies. 2 - Ship Routing Problems: Closing the Optimality Gap Thibaut Vidal, PUC-Rio, Departamento de Informatica, Rua Marques de Sao Vicente, 225, Rio de Janeiro, 22453-900, Brazil, Gabriel Homsi, Rafael Martinelli, Kjetil Fagerholt Hemmati et al. (2014) recently introduced real ship routing instances which pose considerable challenges for solution methods. To face this situation, we introduce a hybrid GA and a branch-price-and-cut-based algorithm. Our GA combines a set- partitioning with problem-tailored variation operators to optimize all decision subsets. Our BCP generates elementary routes and relies on strong branching, sophisticated preprocessing and correction techniques for the delivery triangle inequality, route enumeration and subset-row cuts. As visible in our experiment, the BCP solves to optimality all of the 240 available instances, and the GA finds solutions which are near optimal in a few minutes. 3 - Dynamic Car-passenger Matching for on Demand Mobility Services Marvin Erdmann, University of the German Federal Armed Forces, Munich, Werner-Heisenberg-Weg 39, Neubiberg, 85577, Germany Marvin Erdmann, BMW, Parkring 19, Garching, 85748, Germany Today’s challenges of urban traffic - congestion, lack of space, air pollution, etc. - can be met by On Demand Mobility, a concept that would lead to an enhanced use of shared mobility services to utilize the vehicles more efficiently. This trend becomes even more obvious with the future development of autonomous driving vehicles. To avoid a decline of flexibility and convenience for the costumers the fleet management function will match the requests and the vehicles in order to quickly find a reliable and time efficient solution for the whole system. My work’s focus is the realization of a Tabu Search Metaheuristic to compute near-to-optimal solutions for this NP hard problem. n SC33 North Bldg 222C Strategic Decision Making for Emergency Medical Services Sponsored: Transportation Science & Logistics Sponsored Session Chair: Lavanya Marla, U. of Illinois at Urbana-Champaign, Urbana, IL, 61801, United States 1 - Competitive Coverage with Two EMSs Peter McGlaughlin, University of Illinois-Urbana Champaign, Urbana, IL, United States We study a stochastic game in which two service providers (the players) compete for customers. Player maintain a base at one end point of the service area, a line of length 1, and selects a coverage range 0

n SC31 North Bldg 222A Modeling and Evaluation of Transportation Infrastructure with New Technologies Sponsored: TSL/Intelligent Transportation Systems (ITS) Sponsored Session Chair: Michael Levin, University of Texas, Austin, TX, 78731, United States 1 - A Scalable and Real-time Approach for Optimal Signal Control in a Semi-connected Vehicle Environment Ali Hajbabaie, Washington State University, 405 Spokane St, Pullman, WA, 99164, United States, S. M. A. Bin Al Islam, Husain Aziz This research presents an algorithm to incorporate point detector data with shared information of connected vehicles (CV) for real-time optimal signal control with various CV market penetration rates. Besides, a transit signal priority (TSP) based mixed-integer linear program (MILP) is formulated to accommodate multiple modes of vehicles while considering effective coordination between neighboring intersections. The results indicated significant improvement in network performance under various CV penetrations rates compared to the state-of- practice TSP using microscopic traffic simulation. 2 - Performance Evaluation of Connected and Cooperative Driving: Queueing Theoretic Approach and Case Study Mohammad Motie, University of Southern California, Los Angeles, CA, 90089, United States, Ketan Savla We develop a queueing theoretic framework for performance analysis, in terms of throughput and travel time, for freeway systems under connected vehicles and cooperative driving. In particular, we derive theoretical bounds under well- studied car-following models, and compare with microscopic simulations and the NGSIM dataset. 3 - A Machine Learning Based Signal Control Algorithm with Energy Minimization Objective S. M. A. Bin Al Islam, Graduate Research Assistant, Washington State University, 242, Sloan Hall, NE Spokane St, Pullman, WA, 37830, United States, Husain Aziz, Ali Hajbabaie This paper presents a distributed agent-based reinforcement learning (RL) technique to minimize energy consumption from signalized intersections while considering effective coordination in decision making processes among adjacent intersections. Besides, we propose a mobility-energy objective to provide a balanced reward function in RL based signal control algorithm. The algorithm is implemented in a calibrated NG-SIM network within a traffic microsimulator-PTV VISSIM and compared with current signal settings of the network and hypothetical optimal solutions. 4 - Network Traffic Control for Autonomous and Legacy Vehicles Michael W. Levin, University of Minnesota, 500 Pillsbury Dr. SE, Minneapolis, MN, United States, David Rey Autonomous vehicles are expected to help in improving network capacity and mitigate congestion episodes. Yet, there is a lack of traffic control models capable of managing both legacy and autonomous vehicles. We address this gap by optimizing traffic signal control at the intersection level considering mixed traffic flow. We then embed this formulation into a network-level traffic control model wherein link queue evolution dynamics are stochastic. We propose a distributed network traffic control policy inspired by back-pressure algorithms to maximize network throughput and we prove stability. Joint Session TSL/ICS: Approximate Dynamic Programming and Reinforcement Learning for Routing III Sponsored: TSL/Freight Transportation & Logistics Sponsored Session Chair: Justin Goodson, Saint Louis University, St. Louis, MO, 63108, United States 1 - Dynamic Time Window Allocation and Sizing for Service Routing Justin Goodson, Saint Louis University, John Cook School of Business, Davis-Shaughnessy Hall, St. Louis, MO, 63108, United States, Marlin Wolf Ulmer, Barrett Thomas Firms serving customers at their homes face the challenge of assigning time windows small enough to respect busy schedules, but large enough to provide adequate service. At the time of a request, the firm must provide a window within which service will begin. Accurately estimating arrival times to customers is difficult because requests arrive randomly and must be assigned to employees n SC32 North Bldg 222B

72