SAEVs Transportation Science

This repository provides a simulation and reinforcement learning environment for Shared Autonomous Electric Vehicles (SAEVs).
It supports modeling of fleet operations, trip matching, charging strategies, and vehicle scheduling to study the efficiency and economics of SAEVs in urban transportation systems.

📄 This repository accompanies the following paper:
Fleet Operations of Shared Autonomous Electric Vehicles: Impacts of Charging Infrastructure and Charging Strategy, published in Transportation Science.

---

---

🚀 Features

• Core Simulation (saev/core/)

  • Models vehicles, trips, zones, charging stations, parking, matching, fleet, and model.
  • Central logging in saev/core/log.py.

• Reinforcement Learning (saev/rl/)

  • saev/rl/environment.py provides the gym environment for training.
  • Agents and configs exposed via saev/rl/agents/.

• Visualization Tools (Visualization/)

  • Scripts for analyzing trips, charging, state of charge (SoC), and performance.
  • Generates plots and PDF reports (e.g., profit_RL.pdf, SoC.pdf, CSs.pdf).

• Main Entry

  • main.py — runs the simulation and RL training pipeline.

---

📂 Repository Structure

saev/
  core/
    charging_station.py
    location.py
    parking.py
    trip.py
    vehicle.py
    Zone.py
    Matching.py
    read.py
    log.py
    model_impl.py
    model.py
  rl/
    environment.py
    agents/
      __init__.py
main.py
Visualization/
requirements.txt
data/
  input/
  output/

Set input data directory (optional):

export SAEV_DATA_DIR=$(pwd)/data/input