The mobility space faces significant disruptions today. The emergence of
automation technologies increased data volumes flowing from vehicles, and changes in existing business models are bringing new challenges to fleet
owners and smart mobility providers. Optimizing fleets involves looking ahead seconds, minutes, days, and weeks to determine where to locate idle vehicles to maximize their utilization, which vehicles to dispatch, when to perform maintenance, where to place electric vehicle charging stations, and how many vehicles to put on the road every day and every hour. Fleetonomy is currently working with a variety of companies in the mobility ecosystem to apply its fleet optimization solutions, including demand prediction, fleet redistribution, and smart dispatch, to their existing fleet management services.
Building and training these machine learning models require input from
massive quantities of data, in particular, historical data from actual vehicles. Automotive data must be matched to related datasets such as location, day of the week, weather, traffic, nearby events, and more.
“With rich data on traffic flow in a particular geographic location, Fleetonomy can deeply understand and predict demand for various mobility services such as ride-hailing, car sharing, car rental, vehicle subscription services, and ultimately — autonomous vehicle trips,” says Lior Gerenstein, Co-Founder and CTO at Fleetonomy. “Fleetonomy’s platform can then use these predictions to simulate how such services would behave in that location and manage them more efficiently in real-time.”
Fleetonomy did the initial work on its models using multiple data sources, including trip data gathered from smartphones, public transportation datasets, and by directly integrating connected car data from vehicle fleets. However, some of these data sources have limitations:
• Smartphone data does not provide enough accuracy because it includes data points from people walking, cycling, or taking the bus. It also does not contain exact markers denoting when trips started or ended.
• Public datasets are only available from a few geographies such as New York and Singapore, and the limited number of vehicles (public transportation and taxis) makes it difficult to extend the models to trucks and private cars.
• OEM data can be hard to integrate and use for purposes outside of vehicle operations. “Each OEM has unique systems, datasets, and formats,” Gerenstein explains.
Fleetonomy turned to Otonomo’s automotive data services platform to enrich the scope of the historical data for its models. “As we looked into different methods for getting historical data, we found that Otonomo offered the best opportunity to use data from real vehicles in our models,” Gerenstein says. “It’s easy, and we didn’t have to invest much time to get it to work.” The
fact that Otonomo could provide both individual data points and aggregated data points (trips) saved additional engineering resources.
The Fleetonomy engineering team began by integrating data from a major
metropolitan area in Europe. The process of bringing the data into its model took only three days. “The data from Otonomo was very well structured and formatted, and the API was clean,” Gerenstein reports. The team then built a simulation to show how Fleetonomy predictive models would improve key performance indicators such as vehicle utilization rates, required fleet size, and average waiting time for a ridesharing customer.
“By being able to demonstrate significant KPI differences before and after the simulation uses our model, Fleetonomy helps its potential customers build business cases for deploying our service,” says Israel Duanis, Co-Founder and CEO at Fleetonomy. “Using this tool has great value for us during our sales process.”
The simulation indicates that using OEM data for demand prediction decreases the median wait times for a ride-hailing customer from 330 to 190 seconds, a reduction of more than 40%. Furthermore, Fleetonomy’s simulation, again utilizing OEM data provided by Otonomo, has shown that the same service level can be achieved using five fewer vehicles. These metrics represent a significant reduction in the size of a large fleet.
“Being able to have a single provider for automotive data, so we can integrate once and get coverage all over Europe and the United States, is a huge win for our business,” Gerenstein concludes. “It allows us to enrich our models and provide incremental value to our customers. For us, the value of Otonomo sums to demonstrable measurements, which are shown both in the planning phase using simulations and in real-time with operational services.”
Fleetonomy offers the next generation of data-driven fleet management solutions to plan, optimize and operate smart mobility services.
• Built simulation for predictive models using historical trip data matched to related datasets.
• Proved that its optimization solution improves key performance indicators for fleets.
• Integrated with the Otonomo platform in three days
Fleet management, optimization, and demand prediction
Automotive Data Types
Microsoft Azure compliant
Fleetonomy provides a cloud-based, end-to-end fleet management solutions designed to maximize fleet operation efficiency, reduce operational costs, and create new revenue streams based on new intelligent mobility services. Combining AI, machine learning, and big data algorithms, Fleetonomy’s platform provides actionable insights that help fleet owners make better day-to-day decisions regarding inventory management, customer preferences, maintenance and more.
Otonomo paves the way for new apps and services that make transportation safer, more convenient, and truly rewarding. We offer the first neutral automotive data services platform, which provides simple, secure data access and transforms data into actionable insights for services such as predictive maintenance, emergency services, on-demand fueling, insurance, and smart cities.