Core Concepts
Leveraging the Eclipse MOSAIC simulation framework to model and analyze the traffic and environmental impacts of different rebalancing strategies for ride-hailing services.
Abstract
The paper presents a simulation-based approach to model and analyze ride-hailing services using the Eclipse MOSAIC framework. The key aspects are:
Input data analysis:
Examined logbook data from a middle-sized ride-hailing fleet in Berlin, Germany.
Analyzed the different scenarios that occur after a ride is completed, such as accepting a follow-up order or returning to the point-of-business.
Identified that return mileage plays a significant role, accounting for over 30% of the total mileage.
Ride-hailing model:
Implemented a ride-hailing model using the Application Simulator in Eclipse MOSAIC.
Modeled the dynamic nature of ride-hailing, including ride order generation, vehicle assignment, and route planning.
Integrated the model with the large-scale Berlin traffic scenario (BeST) to capture the impact on overall urban traffic.
Experiments:
Defined three rebalancing strategies for drivers after completing a ride: return to point-of-business (baseline), wait at drop-off location, and drive to a nearby hotspot.
Generated artificial logbooks based on the input data to overcome limitations of the original dataset.
Conducted a simulation study to analyze the traffic and environmental impacts of the different rebalancing strategies.
Results:
The Wait and Hotspot strategies showed significant potential for reducing total mileage (up to 24% and 18% respectively) and CO2 emissions (up to 25% and 18% respectively) compared to the baseline.
Extrapolated the findings to the entire ride-hailing fleet in Berlin, estimating potential annual savings of 70 million km in mileage and 7.25 tons of CO2 emissions.
The paper demonstrates the effectiveness of using a simulation-based approach, enabled by the Eclipse MOSAIC framework, to model and analyze the complex dynamics of ride-hailing services and evaluate the impacts of different operational strategies.
Stats
The simulation results show the following key metrics for the different rebalancing strategies on a Wednesday and Saturday:
Wednesday:
Total Mileage:
Return (Baseline): 14,724 km
Wait: 11,127 km (-24%)
Hotspot: 12,060 km (-18%)
Total CO2 Emissions:
Return (Baseline): 1,519 kg
Wait: 1,137 kg (-25%)
Hotspot: 1,245 kg (-18%)
Saturday:
Total Mileage:
Return (Baseline): 23,037 km
Wait: 18,354 km (-20%)
Hotspot: 19,739 km (-14%)
Total CO2 Emissions:
Return (Baseline): 2,382 kg
Wait: 1,883 kg (-21%)
Hotspot: 2,038 kg (-14%)