INFORMS 2021 Program Book

INFORMS Anaheim 2021

TC43

3 - Fleet Size and Charging Infrastructure Capacity for Ridehailing Services Using Autonomous EVS Diwas Paudel, University of South Florida, Tampa, FL, United States, Tapas K. Das We develop a robust stochastic mixed integer model which, for any given part (percentage) of the current demands in the city that the SAEV fleet might intend to serve, yields optimal SAEV actions and corresponding capacity plan (fleet size and charging infrastructure) that maximizes gross profit. The model solution yields optimal capacity plan, which varies depending on the decision makers choice of the level of conservatism towards robustness. Expansion of the existing capacity to meet an increased demand and sensitivities of battery capacity and power network configurations are also explored. 4 - Efficient Carpooling and Toll Pricing for Autonomous Transportation Manxi Wu, Massachusetts Institute of Technology, Cambridge, MA, United States, Saurabh Amin, Patrick Jaillet How can autonomous transportation technology be utilized to reduce road congestion? We analyze a carpooling market, where a transportation authority sets tolls on road segments, and riders are incentivized to organize autonomous carpooled trips and split toll prices. We characterize sufficient conditions on the network topology and riders’ heterogeneous utilities, under which the market equilibrium implements a socially optimal trip assignment. We also propose an algorithm for computing the market equilibrium. These results enable efficient implementation of market-based autonomous carpooling services. TC43 CC Room 213A In Person: Innovation/ Entrepreneurship Contributed Session Chair: Marc Eulerich, University Duisburg-Essen, Hagen, 58099, Germany 1 - An Empirical Assessment of the Role of Multinational vs Entrepreneurial Firms in the Development of Markets: Evidence From African Mobile Telecommunications Industry Mohammad Jahanbakht, Assistant Professor, University of Texas at Arlington, Fort Worth, TX, United States, Romel Mostafa, Soheil Hooshangi Although the impact of pre-entry knowledge on firms’ innovative decisions and performance is known in the industry evolution literature, the nature of relationship between firm-level innovations and market-level competitive outcomes is not well-established. Using GMM method with instrument on a panel data from African mobile telecom industry, we find that a small number of firms are capable of implementing trailblazing strategies which result in a disproportionally large impact on evolution of market-level outcomes, such as adoption, price, and industry capital expenditure. We discuss that a differentiating attribute of these firms is their superior pre-entry knowledge of local markets. 2 - Fundamental Limits of Learning: A Mathematical Framework Yian Yin, Northwestern University, Evanston, IL, United States, Dashun Wang A key aspect in human activities concerns how one learns from past experience. The learning curve literature has documented a robust relationship between experience n and unit cost c: cn ~ n- , with [0,1]. Yet rich empirical results across industries have consistently reported a typical ≈ 0.32 and lack of high > 0.5, raising a fundamental paradox: Is the limit = 1 achievable? Here we develop a general learning model and prove a fundamental limit of learning ≤ ≤ 0.5. We further show that both the technology landscape and strategic explorations of new technology are critical for achieving , which have direct implications for the diagnosis, improvement, and planning of many innovative activities. 3 - Do Elite Innovation Companies Need CSR? An Investigation of The Interactions Between Innovation And Ethical Pay, CSR, and Firm Profits Patti Miles, Associate Professor of Management, University of Maine, Bangor, ME, United States, John N. Angelis Companies that invest in CSR or Ethical Pay usually benefit from improved reputation and legitimization in the eyes of the public, investors, and their employees. However, a company that is highly rated by investors for its innovation may be less likely to need such investments. Using an elite sample of innovation companies created by Clay Christensen (Forbes Top 100 Innovative Companies list), we find that innovative companies are significantly more likely to pay median employees more and be more profitable. However, innovative companies do not necessarily spend more on CSR, and CSR does not successfully mediate the relationship between innovation companies and profits.

TC44 CC Room 213B In Person: Data Mining & Machine Learning in Smart Manufacturing General Session Chair: Neng Fan, University of Arizona, Tucson, AZ, 85721, United States 1 - Robust Tensor Regression Mostafa Reisi Gahrooei, University of Florida, Gainesville, FL, 32608-1047, United States We will extend tensor regression techniques to robust versions and will show their efficacy in detecting and removing outliers through simulations and case studies. 2 - Automated Metrology Planning for 3D Scanning of a Freeform Design Using Bayesian Optimization Zhaohui Geng, Assistant Professor, The University of Texas Rio Grande Valley, Edinburg, TX, United States, Bopaya Bidanda 3D scanning is widely used for the dimension measurements of physical objects with freeform designs. The output point cloud is flexible enough to provide a detailed geometric description for these objects. However, geometric accuracy and precision are still debatable. Uncertainties are ubiquitous in measurement due to many physical factors. This presentation first investigates the geometric and spatial factors that could potentially influence the scanning variability. Their functional relationship is modeled as a ‘black-box’ model, which is later utilized to find the optimal scanning settings for variance reduction. A Bayesian optimization approach is proposed to solve this minimization problem. Case studies are presented to validate the proposed methodology. 3 - A Cloud SDN-enabled Network for Cyber-physical Infrastructure in Smart Manufacturing Systems Lida Haghnegahdar, UNT, Denton, TX, United States To improve predictability, security, and real-time performance in smart manufacturing, Software-defined network (SDN) needs to apply timing and secure criterion and detailed analysis in the software systems and network functions. SDN as a controller can run some applications such as intrusion detection systems (IDS), and central distribution systems (CDS) to monitor systems and devices to detect malicious nodes. This research intends to develop a secure and resilient SDN-based system for a smart manufacturing network. This approach by using controller security procedures will help to detect intrusions and provide reliability within SDN-based smart manufacturing. 4 - Evolutionary Optimization of FAB Dispatch Rule Parameters Harel Yedidsion, Research Scientist, Applied Material, Austin, TX, United States, Derek Adams, Emrah Zarifoglu, David Norman The semiconductor manufacturing process is an NP-hard job shop with re-entry problem. As a result, heuristic dispatch rules are commonly used in practice to schedule lots in semiconductor FABs. The dispatch rule logic has tens of parameters, and its performance relies heavily on the ability to fine-tune the parameter values according to the work-in-progress, and the station availability. In this work, we present a parameter tuning method based on Evolutionary Optimization combined with Simulated Annealing. Our empirical results indicate that the proposed approach outperforms other benchmarks, and can be successfully used to dynamically optimize the dispatch rule’s parameter values.

Tuesday, 12:30PM-1:30PM

Amazon Virtual Room 11 Innovation and Diversity at Amazon Panel Session 1 - Innovation and Diversity at Amazon Mallory Craker, Amazon, Seattle, WA, 98121, United States Five senior leaders who work at Amazon host a panel discussion the importance of innovation and diversity in STEM during INFORMS 2021.

124