Dashcam Video-Driven, Route Distinctive Energy Consumption Pre-Evaluation for Electrifying Public Services Fleets

Abstract

Conventional evaluations prior to electrifying Public Services Vehicles (PSVs) generally provide macro conclusions based on urban layout and fleet magnitude. These inaccurate estimations could further result in recommendations to purchase unsuitable vehicles. Predicting the energy of replacement Electric Vehicles (EVs) using city-designated driving cycle data and geographical information could significantly enhance the accuracy, however, requires vehicle s en-route driving data (cycle) which challenging to obtain before the actually deploy of vehicles. In this paper, taking only dashcam videos as input, through an emergent image processing and recognition technology, textual en-route driving data can be extracted. Thus a vehicle model could simulate the driving of an EV and accurately predict the energy usage of each filmed route. The theories and technology presented in this paper were evaluated using real-life dashcam videos and results demonstrating a negligible error rate while energy requirement from operating an EV on the same route is predicted.