Developing the brain behind the Beryl bike-share scheme

Challenges

One of the biggest costs associated with running a bike-share scheme is making sure the bikes are in the right place at the right time. At present, bike-share scheme operator Beryl spends time and resources positioning bikes where they think they will be needed.

Another significant cost to operators comes from maintaining the bikes and manually checking them for signs of wear or damage.

The challenges from a data science perspective are collating the existing data and then identifying patterns in behaviour, as well as analysing the city’s transport networks.

Bringing all this together to provide information about demand in real time requires state-of-the-art knowledge and data science methods that are not currently commercially available. This is where the input from City, University of London makes a difference.

Background

Beryl is a cycling and technology company that has expanded to provide bike-share schemes. It operates in cities including London, Bournemouth and Hereford.

The company’s bikes are equipped with hardware that collects odometer and telemetry data. Beryl is also developing sensors that can detect atmospheric conditions and wear on the bikes to streamline maintenance.

The knowledge transfer partnership (KTP) between Beryl and City will allow the company to make sense of the data it’s collecting and use it to predict demand across a city.

A data scientist has been recruited to work at Beryl. He is under the supervision of two of the university’s leading experts in data science and modelling: Dr Andrea Baronchelli and Dr Chris Child.

Solution

City’s cutting-edge data science methods, including AI and machine learning, allow Beryl to extract behavioural patterns from large amounts of data.

Mobility analysis and the modelling of human mobility behaviour are other crucial elements in this KTP. Working with Beryl gives the academics involved an opportunity to apply their theoretical research in the real world.

Once the data is collated, they will identify correlations and use this to develop the underlying algorithm for the system. In effect, this will be the brain behind Beryl’s bikes.

For Beryl, this project has the potential to deliver significant cost savings. It will result in more efficient bike positioning and distribution. It will also make maintaining the bikes more efficient, with fewer manual inspections required.

Future plans and benefits

This KTP is still in its early stages, but the plans for the project are ambitious. In January 2020, it will begin working with live trial data. The intention is to analyse the status of a bike-share scheme in London.

The benefits to Beryl’s business are expected to be substantial. But this project provides benefits beyond the company.

Using the hardware and AI technology, the project will be able to model where bikes are going and how people are using the city. Understanding travel behaviour will have value to city planners.

This project also feeds into our teaching. It means students are not only developing theoretical knowledge, but also seeing how this can be applied.

The new knowledge that’s developed through this project will also result in research publications.

Improving the efficiency of bike-share schemes has benefits for communities too. It will encourage more people to use these bikes, leading to environmental and health benefits.