Deep Learning Model Based CO2 Emissions Prediction using Vehicle Telematics Sensors Data

Climate change is one of the greatest environmental hazards of today. Global warming, due to an increase in greenhouse gases has resulted in a continuous global increase in temperature. CO2 continues to be the leading contributor to the greenhouse effect, with transport being a major CO2 emission source. The majority of transport emissions are from road transport i.e. vehicular emissions. To control vehicular emission, first, an efficient emission monitoring system is required. Direct sensor installation in individual vehicles is neither cost-effective nor the data is easy to collect. In this paper, a scalable vehicle CO2 emission prediction model is proposed which uses vehicle On-Board Diagnostics (OBD-II) port data. The proposed system uses real-time in-vehicle sensor data to estimate CO2 emission of the vehicle using a Recurrent neural network (RNN) based Long short-term memory(LSTM) model. OBD-II dongles can be used to easily transmit the vehicles sensor data to the cloud, where the LSTM model uses this data to estimate the real-time CO2 emission of the vehicle. The proposed model provides a scalable and efficient system to monitor emissions at a vehicular level. The proposed model has been evaluated using public OBD-II dataset as reported in literature.