Prior to the acquisition of an electric vehicle, pre-evaluation of vehicle energy use is desirable to assess whether the intrinsic vehicle electrical storage capability is satisfactory. However, inconsistency in general vehicle modelling may provide unreliable predictions concerning energy usage. To increase the prediction reliability, the use of route-specific driving cycle data is essential.This paper presents a case study of a novel method of extracting vehicle telemetry data from archived dashcam videos without the need to deploy conventional telemetry techniques. Utilising dashcam videos as input, and employing image processing and recognition technology, textual en-route driving data embedded in the video can be extracted. This data can then, in-turn, be used to model the performance of the vehicle, or an electric equivalent in terms of energy use and emissions. Results from preliminary testing with real-life dashcam videos, demonstrate negligible errors with regards to energy requirements and pollutants emitted from an EV operating on the modelled routes. Consequently, the proposed solution opens up the possibility to gather a significant amount of new data in order to better assess the transport sector’s energy requirements. This is especially important for situations where conventional telemetry is difficult to obtain. In addition, results from vehicle fleet modelling may inform policy decisions with regard to the impact of introducing low emission zones.
Read full abstract