Electric Taxi Charging
I worked on this project to facilitate Havn, a start-up chauffeur service operating a fleet of all-electric Jaguar I-PACEs, with their vehicle charging strategy. The fleet of I-PACEs were usually charged overnight at a low-cost charging station, but would often require topping up during the day to complete their shifts. The company wanted to know whether it would be more efficient for the vehicles to return to the overnight charging station or to use a more expensive public rapid charging point nearer potential customer pick-ups for these top ups.
Each vehicle was fitted with a logging device that would report live location and battery charge data to an AWS server managed by Synaptiv, a start-up company that specialises in making connected car data more accessible. I combined data from this server with ridership statistics from the company’s ride-hailing platform and public charging point information from the Zap-Map.com API to create a model for determining optimal charging times and locations.
I used machine learning techniques such as clustering analysis to categorise the locations and frequencies that public charging points had been used in the past and how this usage correlated with ridership. By comparing this past information with the results of my model, I discovered that the company could save hundreds of pounds per month in charging costs while increasing availability for its customers by choosing appropriate charging points.
The final stage of the project was to implement an alert system that informed the fleet manager if there was an opportunity for a vehicle to recharge during the day. The fleet manager could then use a dashboard to verify vehicle locations in relation to the suggested charging point and upcoming scheduled rides.
