Diesel and petrol vehicles are significant contributors to poor air quality in the UK and globally. There is clear evidence demonstrating the negative effects of poor air quality on human health and the environment.
The Eco Road Warriors took on the task of using data to demonstrate just how much impact drivers could have in reducing their emissions by slight changes to their route. While GPS navigation offers routes that minimise travel time, the focus here is on creating routes that reduce journey emissions in exchange for a modest increase in journey times.
The team modelled likely emissions for each road in the city taking into account factors including traffic lights and crossings. They used as their basis Nitrogen Oxide (NOx) emissions data from the company Cambustion. This data was used to train a machine learning model to predict emissions.
This project has not been without its challenges. At first the team found that their model was not successful as data from roads alone were not useful for emissions prediction. This is when the team then focused on vehicle telemetry (e.g. GPS location, speed, etc.) which gives a more accurate predictive model for NOx emissions.
In addition, while the route planning worked well as proof of concept, as the project developed, it was clear that it would not be able to handle future requirements. This is when the team adopted a well established open source project. This is ongoing.
An Android App has been created. It features a map centred on the current location. This provides an interface searching for origin and destination. The user can than select from multiple possible routes. To note that turn-by-turn navigation is not yet integrated into the system, and work is ongoing.
The team’s current focus is on the vehicle telemetry data and they will be undertaking more modelling; this will entail data collection, initially from Mobilized Construction. Mobilized Construction also pitched, and won, a later round of funding from Bristol City Council under Our Data projects – you can read more about this project here.
The team is anticipating a campaign launch offering interested drivers the opportunity to participate in App trials and provide feedback. This initial launch of the App is scheduled for September 2019.
Project team info:
- Ercan Ezin is a PhD student at the University of Bristol currently working on Context-Aware Recommender Systems for Leisure Planning in Smart Cities. Ercan has 3 years of work experience and has worked on Mobile and Web development. More info at www.ercanezin.com.
- Frank Kelly is a lead data scientist working in the domain of smart cities with over 15 years industrial experience. He is a Cambridge alumni with a background in systems engineering and signal processing; his skill set includes machine learning, agile development and project management. In his spare time he plays squash, snowboards, and runs a community tech meet-up.
- Alexander Frenzel is a python developer with over 10 years of experience, who graduated in computer science (computer vision, neural networks). His expertise ranges from web development, machine learning to Linux system administration. (E-Mail: firstname.lastname@example.org)