AI-based range prediction for EVs proven in trial

Spark EV Technology has completed a trial with BP of its AI-based range prediction system for electric vehicles.

By delivering more accurate range predictions, based on actual driving style, topography and current atmospheric conditions, Spark’s technology overcomes range anxiety. As Spark’s AI algorithms learn with every journey, it continually improves the accuracy of personalised journey predictions.

The project saw BP employees based at five locations drive 10,000 km in two types of EVs equipped with Spark’s journey prediction solution and supplied by Zipcar and Avis Budget Group. As part of the testing phase, drivers entered their proposed journey into Spark’s smartphone application to record data and make comparisons with the onboard range display. They then received personalised advice on whether they could complete it - based on live data, driving style, urban/country routes, previous trips and charge point locations.

During Spark’s trial with BP it was observed that one car could have driven 45 miles more than its displayed range, and on another journey an extra 21 miles more range had to be used to complete the journey than the car predicted. In comparison, Spark intelligent range prediction uses data on where the car is going, not where it has been, and for the same journeys predicted the range to an accuracy of just 3.8 miles and 0.4 miles respectively.

“Our trial with Spark EV helped to increase our understanding of how technology could help reduce consumer range anxiety, especially for new EV drivers,” said Sophia Nadur, Innovation Director, BP. “We now look forward to working together to explore further potential applications into future software iterations.”

 The project began as part of BP and RocketSpace’s Mobility Tech Innovation Collaborative programme, which brings together the world’s most promising startups and global industry leaders to work together on paid projects. Previous RocketSpace alumni include Spotify and Uber. Spark was one of just four companies selected globally for the programme, which attracted hundreds of applicants.