April 2021
“Image sensors are key to vehicle safety”
“We should not forget the lesson that this pandemic is teaching us to reduce the use of private vehicles”
“Interest in researching mobility topics is increasing”
Elisa Sayrol, PhD holder in Telecommunications Engineering, is a researcher specialised in image and video processing. She is associate professor at the UPC Barcelona School of Telecommunications Engineering (ETSETB), where she was the director for six years, and she coordinates the master’s degree in Urban Mobility. Since 2017, she has formed part of the academic management of CARNET, the mobility hub promoted by the UPC, SEAT and Volkswagen.
How did you come to focus your research on the field of mobility?
I arrived at CARNET in 2017 thanks to the academic director of the centre, Lluís Jofre. I came from several years of intensive university management at the UPC, where I was director of the Barcelona School of Telecommunications and Vice-Rector for Institutional Relations, and I gained experience in the development of strategic projects. Moreover, the CARNET academic team includes lecturers from a range of disciplines, including telecommunications. Telecommunications are very important in urban mobility, so that vehicles can connect with the other elements in the system. However, my specialisation is image processing, a line of work that converges with the work of computer engineers in topics of computer vision. And all of this forms a considerable part of the development of autonomous vehicles.
Before you joined CARNET, had you applied your vision research to mobility?
No. When I left management, the use of deep learning was beginning to be extended to image processing, and this got me interested in applying new detection algorithms in other areas, but joining CARNET is what led to me to focus on mobility.
What role do image processing technologies play in mobility?
Driver-assistance systems have advanced greatly. Even in cases in which driving is not fully autonomous, vehicles have utilities that detect what is happening around them, with detection and communication sensors. These include vision cameras and Lidar sensors, among others. A car being able to “see” what is happening around it improves safety and driving decision-making. In addition, communicating the information that it captures about other vehicles and infrastructure is very important. In fact, in the master’s degree in Automotive Engineering at the UPC, a specialisation has been created in connected vehicles and assisted driving that includes subjects on ICT applied to automotive engineering.
This leads us to Artificial Vision for Kick Scooters, one of the research projects that you have worked on recently. What does it consist of?
We are working in the field of urban mobility from the perspective of scooters. That is why we have created a database of videos that include journeys by scooter, which is also valid for bicycles, through various areas of the city. Most of the journeys are on lanes specifically for this use. Subsequently, we label the data, that is, we mark what we want the algorithm to detect and we use the data to train the algorithms. Finally, we test them in new videos to assess their efficacy. The final aim is to improve safety. We have used images to study how scooters face risk situations with pedestrians and other vehicles, depending on the type of route they are moving along. This project has been approached as a test of concept, and now we are proposing a project for EIT Urban Mobility to start a demo that brings us closer to generating a product. In addition, we currently have several students who are researching multi-task algorithms, which enable us to use the same neural network for various functions.
The use of scooters, skateboards and bicycles has increased greatly. How should cities such as Barcelona adapt to this new model of mobility?
A first regulation on personal mobility vehicles has been introduced at state level. Local administrations develop regulations that must evolve to promote safety, but they must also make a physical change to adapt transport infrastructure. In Barcelona, the bike lane network has been extended greatly, but this is not enough. We also need to think about, for example, the width of the lanes so that they are safe and have greater capacity.
Does the image processing in which you work contribute to improving the safety of urban mobility?
Yes. Image sensors are key to vehicle safety. With the development of 5G, along with IoT, they will become even more important.
In research on mobility, is collaboration with companies in the production sector important?
Totally, and this is the role that is being carried out by CARNET. Connection with the problems and challenges of companies enables us to focus the research and provide answers for the production sector.
And with the regulators?
As well. Collaboration platforms need to be established between researchers, regulators and companies to make progress. And in our case there is a fourth element, which is the city as a supplier of services to the people.
How is COVID altering mobility in the mid-term in cities such as Barcelona?
Micromobility and sharing of this type of devices is benefitting [from COVID]. Users are opting for individual, cheap, environmentally friendly means of transport. In contrast, the use of public transport is dropping and the use of individual vehicles is increasing. We should not forget the lesson that this pandemic is teaching us to reduce the use of private vehicles. In this respect, teleworking has helped and we should include it as an option, even if not at 100%.
How was the UPC master’s degree in Urban Mobility, which you coordinate, created?
It was created in response to our participation in the EIT Urban Mobility consortium. The master’s degree is based on another UPC programme. The new programme focuses particularly on boosting innovation and entrepreneurship. In fact, 25% of the credits are for subjects in this area. In addition, the programme is multidisciplinary. To the logistics and transport section is added a section on data management and a section on mobility technologies that address aspects relating to energy, sensors and human machine interaction. We are completing the first edition. This is an international programme that is taught in English and most students are from other countries. My impression is that interest in training and research in mobility topics is increasing.
CARNET and EIT promote hackathons and collaborative research activities on mobility. What is the aim?
We use them to encourage entrepreneurship among new graduates. Interesting ideas emerge from the hackathons that sometimes lead to a business plan and the creation of start-ups. And this enables us to find talent.
Engineer, doctor and director of an engineering school. What should be done to encourage women to enter STEM careers?
I think that the low percentage of female students is related to the lack of role models. If girls knew of successful female engineers, the number entering this field would increase. CARNET is a good example in this respect, because Monika Bachofner and Laia Pagès were adventurous to lead an initiative like this from its outset. In addition, much can be attained by making the social part of engineering more visible, the application that everything we do has and how society benefits from our work. In CARNET, we participate in an EIT project Women in urban mobility in which we address the perspective of gender in mobility. On 12 May we are holding an event in Barcelona. Anyone interested in the topic is welcome to participate.
The article was jointly funded by the European Regional Development Funds (ERDF) of the European Union, as part of the Operational Framework of the ERDF in Catalonia 2014-2020.