Design of Smart Last-Mile Logistics for on-Demand Delivery (Toters-Style)

Design, implement, and evaluate an end‑to‑end decision system for an on‑demand delivery platform operating in a dense city (e.g., Beirut). The system must plan courier capacity ahead of time and optimize real‑time order assignment and routing under uncertain demand, restaurant prep times, and traffic. Students will build a discrete‑event simulator, compare optimization/learning methods, and deliver a reproducible artifact with managerial insights.

This project requires the following main tasks:

  1. Project setup & data model: define zones, entities: orders / riders / restaurants, and Key Performance Indicators (KPIs)..
  2. Synthetic data generator: orders, prep-time, and time-of-day travel profiles.
  3. Forecasting module: demand & prep-time forecasts with prediction intervals.
  4. Staffing optimization (using a MILP): shift templates, vehicle mix, chance/penalty for under-coverage.
  5. Simulation: event-driven engine integrating forecasts, staffing, and dispatch.
  6. Dispatch & routing: rolling a “Pickup and Delivery Problem with Time Windows” (PDPTW) as a baseline using greedy/OR-Tools, then using an advanced heuristic, e.g., Adaptive Large Neighborhood Search (ALNS).
  7. KPI dashboard & logging: define Service Level Agreement (SLA), utilization, cost/order, fairness…

Project Details

[photo]