As the popular saying goes, nothing in life is a coincidence. The same could be said by Luis Rita, a data scientist studying for a PhD in London, and Miguel Peliteiro, a doctor at Santarem Hospital, who have joined forces to develop the CycleAI project, an app designed to identify safe routes for cyclists with the assistance of artificial intelligence.
The two hadn´t met in person until last December, and the first contact between them took place remotely via Skype, in September, after the academic project developed by Luís at Imperial College London caught Miguel's attention. “I don't believe in totally fortuitous circumstances, arising from luck or bad luck”, reveals the doctor to Energiser, recalling that he was undergoing treatment in a hyperbaric chamber after a serious bicycle accident he was involved in a few months before, when he saw a programme on television highlighting the “Top 25 Under-25 Talents” award, won by the data scientist. “That hour and a half was the only free time I had in my daily routine, as I was also studying and undergoing other treatment, and I ended up crossing Luis´ path”, he adds.
The idea, which began as a purely social initiative and as support for people who use the bicycle as a means of transport, is gradually turning into a business adventure that has drawn attention in several entrepreneurship competitions and which has already taken home a series of prizes. The most recent, awarded at the “Hackathon Building the Future” event, promoted by Microsoft with the support of Galp, guaranteed them a prize of 3,000 euros, which will be invested in the next phases of the development of the software. “We have been working on this project on a daily basis and we are now all dedicating more time to CycleAI than to the activities we were conducting before, although this wasn´t the original plan”, says Luis Rita. However, the team, which has grown in the meantime and already has two more members in addition to the co-mentors - Tudor Bostan, a web developer, and Gonçalo Moreno, a machine learning engineer -, is complete for the time being. “We now have a full team to develop the technological solution, and we are capable of achieving exactly what we set out to do”, he explains. With regard to the software, “we will create a tool capable of creating perceived safety maps, while the hardware side of the project will consist of installing cameras and everything else in these vehicles (bicycles and scooters)”, he adds.
Facilitating the life of cyclists
It was while concluding his Master's degree in London that Luis Rita first thought of developing a software programme designed to identify the areas of greatest risk to cyclists, not least because he himself uses this means of transport to get around on a daily basis. The idea was initially limited to the English city, but it quickly gained in momentum and size. “We weren´t worried about making any money from it”, says Luis Rita. However, the desire to do something in the long term and dedicate themselves on a full-time basis forced them to set their sights a little higher because, he says, “achieving this goal wouldn´t be possible without customers and a marketable product”.
For now, the team is working on contacting potential customers and investors, but they want to guarantee their independence so as not to have to rely on third parties. Furthermore, they need partners – basically local government authorities and city councils - for the implementation and use of the safety maps they are creating.
Artificial intelligence makes the difference
Crowdsourcing has been the means of collecting all the information required to identify dangerous areas for cyclists, in several cities around the world, with a view to creating safety maps. The contributions of users registered on the platform (on the CycleAI website) also help ‘train’ the tool. In other words, on the one hand, the user points out the places he regards as risky on the map, providing his reasons and with the option of placing an image and, on the other hand, selecting which place seems to be the safest from a cyclist´s point of view from a series of images. “Up to now, we have identified almost 100 points in 8 or 9 countries, which is already enough to validate our tool”, explains Luis Rita. With regard to the identification of images, CycleAI has almost 5,000 participants and more than 40,000 votes from 36 different countries.
In this context, and in possession of this data, artificial intelligence will make the difference. Finding patterns in thousands of pieces of data and ‘learning’ from them will be the way in which to put into use all the information on the safety map provided by users.
Asked about how they envisage the project in a year´s time, Miguel and Luis are unanimous: “we want to have our own smart vehicles in operation and to reduce the current number of accidents involving scooters and bicycles by over 90%”.