Meet Veronica Marin, our advanced Algorithms expert!
Can you tell us about the work you are doing?
Right now, I lead the advanced algorithms team of the Research & Technology Department here in Toronto, Canada. We are working on what we call vehicle situational awareness. There are many applications in railways for AI and we are using our expertise to support global ground transportation projects in urban and main line rail.
What is vehicle situational awareness?
It means bringing smart capabilities to each train. The train needs to determine what is happening around it and to react appropriately. For example, the train needs to be aware when it is approaching a station with a crowded platform that does not have platform screen doors. It’s about adding new layers of safety on top of the automatic safety functions that are already part of CBTC signalling – it does not replace existing safety systems.
How can trains see?
We use different types of sensors: like for example cameras or radars.
This is the same sensor suite used by the driverless automotive industry. Tesla, Mercedes and BMW also use sensors of this type. Sensors make it possible for a vehicle to “see” its surroundings and to capture data about objects around it.
Does this mean trains need to think as well?
Yes. The first step is object detection. This involves classifying objects, and then applying high-level functionality to replicate perception – similar to human decision making. We achieve this with advanced algorithms. We need to simplify all of this into a rules-based system, and to build the system based on the rules we learn as humans.
Will trains be able to learn for themselves?
This is what we envisage for the future. But in the first instance, we are going to train the systems offline: we collect data, we train it, then we deploy. That way, we have a full understanding of what is happening. Explain ability is key – you need to pinpoint how a decision was made, and know what led the algorithm to output that decision. The key for us is to identify how we can gradually take advantage of machine learning and AI without compromising the safety behind the systems we are deploying.
What is the prize here?
It’s about being able to address rail industry pain points using AI, and identifying a strategic action plan to make this real. That’s my day-to-day role right now. We start at the level of the vehicle – the train. For example, you can adapt the way the train is driven based on track conditions to reduce energy consumption. There are also benefits at system level: the technology enables correlation between multiple variables, so it is possible to orchestrate operations much more efficiently.
How important is collaboration in developing AI applications?
It’s very important. We are sponsors of the Vector Institute for AI in Toronto. Access to academic thinking is valuable, particularly when it comes to ethical questions, or evolving areas such as symbolic AI. In addition, we are an active member of study groups and we are engaged in the development of standards. We are in the IEEE and we are always thinking about how we can better shape the next version of the standard. We also look for partners: it could be in safety-critical systems, certification, or showcasing new capabilities. We work with small and medium-sized companies and identify things that can be outsourced to accelerate our roadmap.
What academic background do people need to work in your department?
Courses related to computer vision, machine learning and deep learning are valuable. Knowledge of supervised learning and unsupervised learning is also important, as well as mathematics and advanced statistics. It’s even better if this is paired with a capstone project based on a real-world scenario. The work we do is interdisciplinary: the stronger candidates have hands-on experience of integration across different knowledge areas. When we interview, we establish what stage of the journey the candidate has reached. We then identify needs within the team and create opportunities so they can make a real contribution.
Are there enough women in the rail sector?
There is still a long way to go! I come from Venezuela. We are an oil-producing country, and that attracts a lot of women into technology and engineering. When I came to Canada, I was surprised to find that women were not being attracted to these roles. The first thing we need to do is to make sure that young women understand there is a place for them. We want to hear their voices because they bring value. This is especially important in the work we are doing. When I have had women in my team, the dynamics for creativity changed a lot – you get stronger ideas. That said, diversity is one of the strengths of our company. I work with people from Singapore, Germany – just about everywhere. There are no boundaries and we really make it work as a team.
What attracted you to the rail industry?
Helping people is an important part of my life. Rail engineering is a great way to do that. People rely on railways to get safely from A to B. Rail also plays an essential part in creating a more sustainable world – we can’t all drive around in cars. Reducing the impact on the environment is something that I care about a lot. So helping people and creating a better world are the two things that attracted me to the rail industry.
What’s your favourite train trip?
When I first came to Canada, I stayed close to Niagara Falls. My journey to work involved commuting to Toronto on the GO Train along the shore of Lake Ontario. This is my favourite memory of coming to Canada – whenever I see that train, it reminds me of that time. It’s like a trip down memory lane.