CTO Aleks Zoranic discusses BusPatrol’s kids first data-driven innovation for school buses and the technical challenges of modernizing thousands of school buses through data driven solutions.
BusPatrol’s CTO has been here from the very beginning. From the outset he’s been dedicated to a kids first data driven innovation for school buses. It’s hard to believe when speaking to Aleks Zoranic that he has 15+ years of experience in the student safety space. He looks like a recent university graduate, however his resume tells a profoundly different story.
Staying young and current is an essential part of his role and a necessity that the job demands; remaining at the cutting edge of IoT and developing teams where innovation and improvisation are constant. This energy paired with his experience allowed me to get a glimpse into the culture of BusPatrol and how it facilitates the mission of upgrading every school bus in North America.
Based in Louisiana, Aleks started developing AlertBus during his graduate studies at the University of New Orleans. He joined the team when there were just two other people in the office. With his background as a Coder, Aleks was thrust into developing the Front-End and Back-End Edge device platform. Aleks started from the ground up building and customizing the computer hardware that he would install on transit buses and eventually school buses.
These first steps into the unknown would drive his passion for problem solving. Now that we have this hardware and are developing the software to enhance student safety…how many buses can we scale? How can we make this process easier for Police and School Boards? What’s next?
I sat down with Alex to talk about some of the technical challenges that he’s tackled. I also asked about how he managed to help build a culture of problem solving for the kids first data driven innovation.
So Aleks, tell me how you got started working in student safety developing a software that would eventually become AlertBus?
ALEKS: All that was many, many years ago, about 10-11 years…I have to do some math. But it was a call that I received through a school network to jumpstart this idea of child safety platform development. Ended up being two people at the office and with that, I interviewed with the secretary, I interviewed with the Operations Manager at the time, and there was one other person at that office. It was a really small environment, a really small office, and I got the job right away and I had the feeling that in the first few seconds of our interview that they liked me. And with that, because I was working on my graduate studies at University of New Orleans, I would also teach some Computer Science courses.
So over the next few years, I’ve managed to snatch some of my best students and come join BusPatrol. And since then, we’ve been just growing in the first few years, mainly with students from University of New Orleans. And then we expanded that search to every other part of the country, and here we are.
Introducing CTO Aleks Zoranic and our kids first Data Driven approach
Can you describe some of the problems or stop-arm violations challenges that you faced, especially working on a new product in a small shop?
There were definitely different sets of challenges back then than we have today. Of course, in the beginning stages of company growth there’s a lot of adventure, a lot of unknown, uncharted territory. So we were just toying with the idea of deploying some homemade computers on transit buses, possibly on school buses. And then what can we do in that domain? How can we make the transit bus and the trolley bus ride more enjoyable and safer?
How can we make kids’ lives safer while commuting from schools to their homes, from homes to their schools? (This is the basis for Detective Kids First Data Driven Innovation for School Buses) And that’s where we kind of started coming up with different ideas. I mean, you can see how easy it is to come up with solutions to the existing problems that the riders are facing. Back then it was both children on school buses and riders on transit buses.
We all have taken a school bus, right? I’ve taken a school bus. I lived in a very remote neighborhood where a lot of kids from the neighborhood were taking a school bus. And it was a very stressful experience both for students, for parents, for bus drivers, and for schools. Once you get on that bus, you look out of the window and you see people trying to go around the bus as much as possible, right? Those buses are moving slower than any other traffic, they’re cutting off the bus. When the bus stopped, they tried to go around the bus on the left and on the right side of the bus. Pay attention to, you know, the big, yellow, clunk of metal that is transporting these kids with brightly flashing red lights with a stop-arm. And then inside of the bus, you know, it’s like any school: kids are interacting in one way or the other. They’ll exchange words, they’ll sometimes exchange objects that fly into your direction. And I’ve gone through a few bus drivers. Some of them are nice. Some of them are not nice. Some of them, I remember, would leave me at the wrong location that wasn’t my home. Being new to the neighborhood I got lost and it was also pouring rain. At the time, nobody could prove that. There was no technology that would track that bus. There was no technology that would surveillance the kids inside of the bus or surveillance the traffic outside of the bus. Everything that happened just happened. Sometimes the bus driver would get extremely mad, they would stop the bus, get out of the bus and start yelling at the traffic. But that only warns a few people, right? Sometimes they’ll stop the bus and try to calm down the kids on the bus. The commute is not always the most fun part of your day. Kids tend to be very distracted and try to entertain themselves and you know, try to come up with stuff to do.
Based on what we’ve discussed in regards to kids first data driven innovation, I see that seamlessly bridging hardware and software is critical. What type of languages do your Programmers and Software Engineers use and how has that evolved over time?
In the beginning, we chose the language that was the easiest and quickest for us to write. And, of course, those were Windows based languages. Since then, we moved to an embedded platform, started using embedded Linux, and some more powerful back end technologies of Amazon Web Services. We’ve progressed into C, C++, Python. On the backend, there’s a lot of Ruby on Rails. There’s a lot of Python coding for the artificial intelligence that powers our analytics. And there’s a lot of AWS services that we utilize to bring our solution to what it is now.
Let’s zone in on the AI for a second. I know BusPatrol is looking for Data Annotators and, you know, Video Reviewers on that side. Will their work be able to train the AI? How does that work?
There’s a lot of data that now comes from our Edge devices. There’s a lot of audio that we tap into on the bus. There are cameras with both video and audio data, GPS data, environmental data. We store all the data in our databases, and we store most of that video on the Edge device until it’s transferred to our back end.
We’re not even doing everything that we could do with the data, especially with the video data. So what we have been doing for the past few years is we’ve been using that video data to train our models to detect…to recognize what’s going on, on the outside of the bus. So what defines erratic driver behavior to the exterior of the bus? What constitutes a violation? A driver trying to go pass a deployed stop arm, speeds that they’re going at, right? So that’s just outside of the bus. Inside of the bus, what can we detect? The things that I’ve been facing, that I was facing when I was taking the school bus was bullying, you know, somebody brought scissors or a utility knife or just a regular knife on board of school bus. We can use that data inside of the bus to observe what’s a normal child behavior and what’s an erratic child behavior, right? Can we recognize weapons? Can we recognize knives, guns? Can we recognize not just objects? Can we recognize actions when somebody is getting into a fight? When somebody is elevating their voice at another person? What constitutes a distracted bus driver, right? Are they using their cell phones? Are they looking elsewhere and not on the road? But also, we can scan the entirety of the interior of the bus and see if there are any children left on the bus after the bus driver parks their vehicle.
You’re creating the entire safety environment and the data informs your approach to scaling that system. How has it been integrating these kids first data driven innovation for school buses into your systems? In addition, you have technology partners so you are integrating diverse hardware and software systems…
We start from the kids first. We look at what technology and systems would increase the safety of children inside of the bus. These are all optional in our program — monitoring bus driver behavior, monitoring the exterior of the bus for possible violations. By working with our partners Zonar, Samsung and then AT&T by leveraging their priority network first, we can assure that the data is transferred as quickly as possible. Timing is always an issue. The video files, when streamed or transferred in 4k definition, are quite large. So what you want to do is, you want to analyze that video on the bus, you want to feed that analytics to the back end, then you want to make an informed decision. And all that has to happen quickly. By using AT&T and FirstNet, we make sure that the data is prioritized on the network, and I’m sure it’s more important than you know, other drivers around that school bus texting while driving. So by having that data traffic prioritized we assure quick analytics and quick communication between the bus and our back end. Then there are various checks that the bus drivers that are mechanics or service field techs do for the bus to make sure that the bus is 100% equipped, that it’s working, that all the connectivity to our device is 100%. And by installing these tablets, we can monitor the statuses of those installations. We can tap into the feed…the video feeds – access the analytics of the bus. We can make sure that the pre-trip and post-trip inspections are conducted by leveraging a relationship with Zonar. Like I said, we store a lot of data, we store a lot of video and GPS data. That data can be used not just by our AI. You can just imagine the possibilities of optimizing routes for these buses.
This interview has been edited for length and clarity
BusPatrol is rapidly growing its team. Work with a cutting-edge AI IoT startup. Specialized and entry-level positions are available with paid training.
Learn more here: www.buspatrol.com/careers