3 présentations

• 
  10h30 - 10h55

  Learning from Driver Experience: a framework for improving vehicle routing

  • Mohammad Hesam Rashidi, prés., University of Toronto
  • Mehdi Nourinejad, York University
  • Matthew Roorda, University of Toronto

  Traditional optimization-based models of the Vehicle Routing Problem are prone to assumptions and simplifications, limiting their applicability to the real world. Data-driven optimization techniques have gained popularity in recent years as they seek to enhance routing quality by leveraging real-world data. Delivery drivers possess tacit knowledge about their daily operating environment, which can be used to develop more effective last-mile planning. For instance, drivers may avoid neighborhoods during peak congestion periods or areas lacking necessary amenities such as parking, which are factors that may be challenging to predict or incorporate into traditional optimization models. This study develops a closed-loop learning framework for the Vehicle Routing Problem, aiming to improve the efficiency of routes for delivering goods to customers by incorporating the expertise of in-field drivers. By leveraging drivers’ experience, the proposed framework provides more robust and high-quality solutions more efficiently. The framework is an extension of existing research on algorithmic deviations and human-centric optimization methods that aim to address the limitations of traditional optimization-based methods in real-world contexts.
• 
  10h55 - 11h20

  Joint Inventory Replenishment and Online Order Allocation Problem in Omni-Channel Retailers

  • Sinem Kinay Savaser, prés., Rotman School of Management
  • Opher Baron, Rotman School of Management, University of Toronto
  • Andre A. Cire, University of Toronto Scarborough

  We consider a strategic problem for an omni-channel retailer with a single distribution center (DC) and multiple stores. We consider walk-in demand at stores as well as online demand from originating from multiple regions. Given these demand streams, the omni-channel retailer simultaneously makes inventory replenishment decisions and allocates online orders either to the DC or pools them at stores with the walk-in demand. We first work on a single-item problem and identify conditions under which a DC is necessary to handle online orders and when the online orders can be shipped from the stores. We provide in-depth numerical experiments to derive managerial insights from these conditions. We also consider capacitated and uncapacitated variants of this problem. We then extend the analysis to multiple item orders and investigate how such online orders affect the order fulfillment decisions.
• 
  11h20 - 11h45

  A consolidated subscription-based delivery program

  • Behnaz Naeimian, prés., York University
  • Mehdi Nourinejad, York University
  • Peter Park, York University

  Same-day delivery has become prevalent due to its convenience and time-saving benefits, but it comes at a high cost due to the premium nature of this service. This study aims to bridge the gap between cost and convenience; therefore, we proposed a consolidated delivery system from pick-up locations to customer doorsteps cost-effectively. The proposed system will operate on specific days in order to be able to combine deliveries efficiently and reduce transportation costs. Using continuous approximation, we estimated the distance required to travel to serve customers. A Profit model is then proposed for a subscription plan for delivery service, and the results demonstrate how profit can vary based on the frequency of delivery services.
  Keywords: Subscription program, last-mile delivery, transportation planning, logistics.