MODEMS: A Tailored Solution for Promoting On-Campus Passenger Mobility
This paper presents MODEMS, a simulation-based framework for on-campus electric ridepooling that formulates the routing challenge as a heterogeneous Electric Autonomous Dial-A-Ride Problem and solves it using a combination of Mixed-Integer Linear Programming models and an Adaptive Large Neighborhood Search algorithm within a dynamic rolling-horizon approach.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine a university campus as a small, self-contained city. People move between buildings, but for those with limited mobility or heavy loads, walking the distance can be difficult. While bicycles and scooters help, they require the rider to be able to operate the vehicle. To solve this, researchers at the University of the Bundeswehr Munich are developing a system where small, electric, driverless shuttles pick up passengers and take them directly to their destination. The challenge is not just building the vehicles, but figuring out how to manage a fleet of them efficiently. These shuttles have limited battery life, they must arrive within specific time windows to be useful, and new requests for rides can pop up at any moment. If the system is too slow to calculate a route, the ride is missed; if it is too rigid, it might run out of power or leave passengers waiting too long. The goal is to create a digital brain that constantly re-evaluates the situation, matching passengers to the right vehicle in real-time while ensuring no one is stranded and no battery dies mid-trip.
In a new study, a team of researchers has built and tested a software framework called MODEMS to solve this exact problem. They treated the campus as a complex puzzle where every passenger request is a piece that must fit into a moving picture. The researchers created three different mathematical models to solve the routing puzzle. The first model was simple and fast but forced the system to accept every single ride request, even if it meant the shuttles would arrive late or run out of power. The second model was stricter, refusing any ride that didn't fit perfectly within the time limits, which sometimes meant rejecting passengers entirely to keep the schedule tight. The third model was a hybrid, allowing the system to be flexible with arrival times and to reject difficult requests if necessary, while carefully tracking the battery levels of every vehicle to ensure they could always return to a charging station. Alongside these precise mathematical models, the team also developed a search method that works like a skilled navigator: it starts with a good plan and then constantly tweaks it, trying thousands of small changes to find a better route without getting stuck in a dead end.
The researchers tested these methods using a digital simulation of their actual campus, which includes a network of roads, designated pickup spots, and a fleet of three different types of electric shuttles. They ran hundreds of simulations to see how the system would perform during a busy workday, with requests arriving at random times and in clusters. The results showed that the strict mathematical models worked well when the number of requests was low, but as the demand grew, they became too slow to calculate a solution in time. The search-based method, however, remained fast and reliable even when the system was flooded with requests. It consistently managed to accept more passengers and used less energy per ride than the other methods. A key finding was that the timing of when the shuttles started their day mattered greatly. If all vehicles started at the same time, they tended to run out of power simultaneously, forcing the system to stop accepting rides while they all recharged. By staggering the start times so that vehicles began their shifts at different intervals, the system maintained a steady flow of service throughout the day, avoiding those periods of total shutdown.
The study also revealed that the way the system handles the "what if" scenarios is crucial. The researchers found that a specific technique for checking if a new ride could fit into an existing route saved a tremendous amount of computing time. Instead of recalculating the entire journey from scratch every time a new passenger was considered, the system could quickly determine if the change was possible by looking at just the relevant parts of the route. This efficiency allowed the system to make decisions in milliseconds, a speed that is essential for a service that operates in real-time. The simulations demonstrated that while a perfect, mathematically proven solution is ideal, it is often too slow for a dynamic environment where conditions change every minute. The flexible, search-based approach proved to be the most robust, capable of adapting to sudden surges in demand and keeping the shuttles moving efficiently.
Ultimately, the work provides a blueprint for how to run a shared, on-demand electric transport service on a campus or in a small city. The researchers showed that by combining flexible routing rules with smart search algorithms, it is possible to keep a fleet of electric vehicles running smoothly without wasting energy or leaving passengers behind. The system they built is currently a proof-of-concept, meaning the software logic has been tested in a simulated environment but not yet deployed on real vehicles. However, the results suggest that such a service could significantly improve mobility for students and staff, particularly those who cannot easily walk long distances. The team plans to move forward with implementing the software on actual vehicles to see how the digital predictions hold up against the unpredictable reality of traffic and weather. Their work highlights that the future of campus mobility lies not just in the vehicles themselves, but in the intelligent, adaptive software that guides them.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.