Investigating the impact of naturalistic driving behaviour differences on energy consumption and road safety

Abstract

The research focus is on two major current challenges facing fleet operators and motor insurance providers globally. For fleet operators, staying profitable means reducing their running cost by increasing their operation efficiency and reducing their running costs by reducing driver fuel consumption and avoiding costly road accidents. For motor insurance providers, having a profitable business requires sale premiums that reflect real-world exposure to risks. This means moving towards metrics beyond traditional pricing parameters, such as age, occupation and gender. Regarding this, there is now substantial interest in understandingdrivers driving behaviour, since knowing which ways of driving are resulting in using morefuel and/or being exposed to road accidents can bring huge financial benefits to both the aforementioned industries.To identify driving behaviours that are costly (both in terms of fuel usage and involvement in road crashes), a naturalistic driving behaviour field study, called Eco Safe Driving Challenge, was developed between 2013 and 2015 at University of Bath. The project followed the same format of major real-world driving behaviour test such as 100-cars in the US, PROLOGUE and UDRIVE in the EU by following the FESTA-V approach in designing the experiment, data collection, data management, ethical and legal concerns, data analysis and evaluating thefindings. In total, 250 km worth of driving data was collected from nine drivers aged between 25 – 30 using medium size petrol cars. A special route was designed to mimic specific road settings, including downhill driving, uphill driving, traffic lights, a pedestrian crossing and a roundabout. It comprised a 4 km loop (same start and end points), starting and finishing at the University of Bath, with an 11% increase and decrease in road slope. The data were collected20 using OBDII dongles fitted with GPS sensors to record location and a sim card for sending driving data to a dedicated server in real-time.A strategic framework has been developed to analyse the data in two domains, i.e. eco-driving and driving and in three phases. First, identifying specific driving behaviour, secondly classifying and comparing drivers differences and finally, scoring, ranking and model drivers’ driving behaviour performances with the aim of assessing eco-driving and safe driving behaviour impacts.The key contributions of the work are, firstly it has found that drivers’ with a tendency to misuse gears, use excessive engine power and/or frequently speed are less fuel efficient than their counterparts To validate this claim, a metric called Vehicle Specific Power – Fuel Consumption (VSP – FC) has been developed, which shows that eco-drivers, on average, have 1.0 – 1.2 points higher than others according to this measure. When evaluating safe drivingbehaviours, it emerged that there was 75% correlation between historical crash zones (based on public records) and locations (400 metre long road segments) where the nine participants in the driving event undertook harsh breaking and acceleration. This provides evidence that scoring drivers based on the number of harsh braking and acceleration events should be included in metrics aimed at evaluating driving behaviour.

Date of Award	1 May 2017
Original language	English
Awarding Institution	University of Bath
Supervisor	Chris Brace (Supervisor)