Electric Ride-Hailing: Balancing Profit and Efficiency

Ride-hailing services like Uber and Lyft are changing how people get around, especially with the rise of electric vehicles (EVs). These services allow passengers to quickly get a ride, while Drivers aim to make a profit. This study looks at how drivers in these services decide when to pick up passengers and when to recharge their electric vehicles.

Drivers face a choice: while they are waiting for a passenger, should they keep looking for rides, or should they go to a Charging station to recharge the battery? This decision-making process can be thought of as a game, where each driver must consider not only their own situation but also what other drivers are doing at the same time.

#The Ride-Hailing Market and Electric Vehicles

In recent years, the number of electric vehicles on the road has skyrocketed. For example, in 2021, global sales of EVs more than doubled compared to the previous year. Many governments are encouraging this shift to electric vehicles by setting targets and investing in charging stations. For instance, the U.S. plans to have half of all new car sales be electric by 2023, while the European Union aims to stop the sale of fossil fuel vehicles by 2035.

Despite the burgeoning market for electric vehicles, there isn’t much research focused on drivers in ride-hailing services. Most studies have either looked at how to plan charging stations or how energy companies manage power generation. This research tends to assume that drivers always need to recharge, but that’s not always the case. Many personal EV owners plug in at home or work, but ride-hailing drivers often travel long distances throughout the day and are more likely to use public charging stations.

#The eRIVER Model

Here, we introduce a new model called eRIVER that focuses on how electric ride-hailing drivers make decisions about picking up passengers and charging their vehicles. The model assumes that every driver is trying to maximize their profit throughout the day. Key to this model is a process that tracks how drivers change their behavior based on the actions of all drivers in the area.

In the eRIVER model, the ride-hailing area is divided into different zones, each with charging stations. The study uses a time frame divided into equal parts, which helps to simplify the analysis. Each charging station can charge a certain number of cars at a time.

As drivers operate, they can be in different states depending on their battery level, location, and whether they are actively searching for passengers or charging. These states help define the available actions for drivers, such as cruising for passengers or heading to a charging station.

#Decision Process for Drivers

Every driver’s decision-making involves several factors, including the number of passengers waiting for rides in their area and how many other drivers are nearby. When looking for passengers, the likelihood of successfully picking someone up is influenced by the number of drivers and the amount of Demand in the area. If a driver can’t find a passenger after a while, they will need to decide whether to keep looking or go to recharge.

When a driver arrives at a charging station, they may find other cars already charging or waiting for a spot. The amount of time they need to wait before charging can vary based on how many cars are already there.

Once a driver makes a choice-whether to pick up a passenger or recharge-they earn rewards or incur costs based on their actions. For instance, picking up a passenger gives a positive return, while charging incurs costs due to the price of electricity. If they run out of battery, they face a significant penalty.

#Mean-Field Game Framework

In this model, we think about every driver not only in isolation but also as part of a larger group. Because the number of vehicles is large, the actions of one driver don’t significantly change the overall situation. Instead, all drivers’ actions together create a general behavior pattern.

The collective behavior of drivers can be seen as a mean-field game, where the average actions of all drivers play a role in defining the best strategy for each individual. This is different from simply solving a standard problem in isolation. Instead, drivers are making decisions that are influenced by the overall flow of traffic and the distribution of other drivers.

#Numerical Analysis

To see how well the eRIVER model works, we conducted experiments based on a simplified version of the ride-hailing system. We looked at how drivers would behave under different demand patterns, such as consistent demand over time, peak demand during rush hours, and concentrated demand in city centers.

In our analysis, we found that a lot of drivers preferred to operate in the central zones, even though these areas had the same general demand as others. This led to high Congestion not only in passenger pick-up areas but also at charging stations. Many drivers would show up to recharge at the same peak times, resulting in long lines and wait times.

#The Issue of Congestion

The study reveals that the competitive behavior of drivers can lead to inefficiencies. Drivers tend to cluster in popular areas, which creates a lack of availability for passengers in less busy zones. Additionally, when cars all rush to charging stations at the same time, it creates bottlenecks and delays.

For instance, many vehicles are idle and unable to pick up passengers during certain times of the day, leading to lost opportunities. The model shows that even minor shifts in how drivers distribute themselves can greatly affect wait times and service quality for passengers.

#Initial Conditions Affecting Performance

Another aspect we explored was how starting conditions for drivers could influence congestion. For example, if all cars begin with fully charged batteries, they tend to run out of power around the same time. This leads to a situation where many vehicles are trying to charge at once, worsening the congestion at charging stations.

We also tested scenarios where drivers started with random battery levels or locations. The findings suggest that varying battery levels have a more significant impact compared to where drivers begin their day. If more drivers start with partially charged batteries, charging queues at stations tend to form sooner.

#Conclusion

This study highlights the complex decisions that electric vehicle drivers in ride-hailing services must make. By using the eRIVER model, we can better understand how these drivers interact with one another and the consequences of their choices on congestion.

As electric vehicle usage continues to grow, it becomes increasingly important to focus on how to balance driver behavior with efficient service for passengers. The insights gathered from this model can help guide future improvements for ride-hailing platforms and charging infrastructure, ultimately leading to a more efficient and effective system for everyone involved.

In summary, as more people embrace electric vehicles and ride-hailing services, there is a clear need for improved understanding and strategies. Drivers will need support in managing their decisions to optimize both their profits and the overall experience for passengers. Future research should also incorporate real-world conditions like traffic patterns and examine how these networks can evolve to accommodate the growing demand for both rides and charging services.