TLDR: CARGO is a new framework that co-optimizes electric vehicle (EV) routing and charging for goods delivery. It addresses challenges like limited battery capacity and charging point availability by integrating delivery route planning with strategic charging decisions. Using a mixed-integer linear programming (MILP) model for exact solutions and a computationally efficient heuristic (CSA) for larger problems, CARGO significantly reduces charging costs (up to 39% over baselines) while maintaining high delivery rates, as validated with real-world data.
Electric vehicles (EVs) are rapidly gaining traction as a sustainable alternative for goods delivery, especially in urban areas. However, their widespread adoption in logistics faces significant hurdles, primarily due to limited battery capacity and the need for efficient recharging. Traditional delivery planning systems, optimized for gasoline-powered vehicles, often fall short when it comes to managing the unique challenges of EVs, such as finding available charging points, managing charging costs, and ensuring vehicles have enough charge to complete their routes within strict delivery timeframes.
To address these complex issues, researchers have developed a new framework called CARGO (Co-Optimization Framework for EV Charging and Routing in Goods Delivery Logistics). This innovative system aims to revolutionize EV-based delivery by simultaneously optimizing both the delivery routes and the charging decisions for electric vehicle fleets. Unlike previous approaches that often tackle routing or charging in isolation, CARGO integrates both aspects, allowing for more coordinated and cost-effective operations.
The core idea behind CARGO is to minimize operational costs while ensuring that all deliveries are completed within their specified time windows. The framework considers various factors, including the availability of charging points, their cost, proximity to the vehicle, and the vehicle’s current state of charge. It also accounts for the time spent on loading, charging, and waiting at charging stations.
The problem of optimizing EV delivery routes and charging is incredibly complex, so much so that it has been proven to be NP-hard, meaning it becomes exponentially harder to solve as the number of deliveries increases. To tackle this, CARGO offers two main solution approaches. The first is an exact method based on Mixed-Integer Linear Programming (MILP), which guarantees an optimal solution but is computationally intensive and best suited for smaller delivery scenarios. For larger, more realistic operations, CARGO provides a computationally efficient heuristic method called Cluster-Sort-Assign (CSA).
The CSA heuristic works by first pre-processing data to identify the nearest and most cost-effective charging options for each delivery location. It then groups deliveries based on their spatial proximity and delivery deadlines, sorting them to ensure timely completion. Finally, it assigns deliveries to the most cost-efficient EV, ensuring the vehicle has enough battery to reach the delivery point and then a nearby charging station if needed. This intelligent approach helps balance the trade-off between travel distance and meeting delivery deadlines.
The effectiveness of CARGO was rigorously tested using real-world data from JD Logistics’ distribution system in Beijing, which includes delivery and pickup requests along with information on 100 charging stations. The results are compelling. When compared to baseline strategies like Earliest Deadline First (EDF) and Nearest Delivery First (NDF), CARGO’s heuristic method demonstrated significant cost savings. It achieved up to a 39% reduction in charging costs compared to EDF and a 22% reduction compared to NDF, all while completing a comparable number of deliveries.
Furthermore, in scalability tests with a larger number of deliveries (up to 200), CARGO’s CSA heuristic proved to be more efficient in terms of execution time than the baseline methods, especially as the problem size grew. This indicates its practical applicability for large-scale delivery systems. The framework also showed that increasing the number of available charging points can further reduce average charging costs, as it provides more opportunities to find nearby, low-cost charging options.
While CARGO represents a significant step forward in optimizing EV-based logistics, the researchers acknowledge areas for future enhancement. These include integrating real-time traffic data to account for congestion, extending the framework to support multiple depots and different types of goods, and incorporating strategies for partial charging instead of full recharges. The implementation of CARGO is publicly available on GitHub, allowing for further research and development. You can find the full research paper here: CARGO Research Paper.
Also Read:
- Dynamic Path Execution for Multi-Agent Systems: The WinkTPG Framework
- Smart Routing for AI at the Edge: Boosting LLM Performance
In conclusion, CARGO offers a robust and practical solution for managing electric vehicle fleets in goods delivery. By co-optimizing routing and charging decisions, it helps logistics companies reduce operational costs, improve delivery efficiency, and contribute to more sustainable urban distribution networks.
