All Articles Infrastructure Transportation How Big Data and AI are helping us get a handle on our traffic woes

How Big Data and AI are helping us get a handle on our traffic woes

Big Data has always been the key to intelligent transportation companies getting better insight into a myriad of traffic issues, and AI has become an important key in improving decision-making and predicting outcomes. But as Jatish Patel at Flow Labs explains in this interview, there are already solutions available today that can make us smarter about transportation.

13 min read


Impact of AI on traffic management

Aleksandar Pasaric/Pexels

Jatish Patel is the founder and CEO of Flow Labs, which he frames as a transportation tech software company specializing in AI and transportation data integration. He’s a serial entrepreneur and represents a new breed of transportation leadership who is fostering the evolution of mobility into the future, and has designed and deployed advanced intelligent transportation technologies. 

As an expert and thought leader in AI for traffic management applications with expertise in machine vision data, processing simulation and optimization, Jatish gave SmartBrief his thoughts on what’s in store for intelligent transportation in the coming year and how AI is fast becoming an important factor in making transportation smarter. 


[Listen to an extended version of this interview, which includes Jatish’s opinion on CV2X and an update on Flow Labs.]


What are your thoughts on intelligent transportation leading into 2024? Where do you see most of the action? And does that align with what technologies are worth advancing right now?

Jatish Patel: We expect it to be a great year coming off the back of 2023, which was a pretty difficult year for a lot of software and data companies in our industry. We saw a couple of major data suppliers in the industry go out of business, which left a lot of transportation solution providers without data to service their clients for about 7 to 8 months and still it’s ongoing.

Whereas 2023 was fairly defensive, 2024 on the offensive side is really going to be a time of growth for the data and software landscape. What we’ve been seeing is that transportation agencies have a better understanding of data and, as well as all of the different things that can be done with it, which enables them to better understand how it can impact their communities today. And they are incredibly eager to invest in traffic signal management, roadway safety and environmental impact which will drive a lot of next year.

Jatish Patel, Flow Labs founder and CEO
Jatish Patel gives SmartBrief some of his opinions on the myriad ways AI will affect the evolution of intelligent transportation.

One of the things that’s also been a huge boon is, over the last few years we’ve seen major investments from the federal government into transportation infrastructure and projects that improve emissions, improve safety, improve mobility. One of the great things that the Federal Highway Administration has done as part of this is actually tied a lot of this grant money with requirements to measure performance. So what that’s doing is driving greater accountability for a transportation agency, which is responsible for investing hundreds of millions of dollars of taxpayer money.

It’s important that companies be accountable in terms of what is the actual impact on the communities that they serve, so we’re seeing a greater focus on performance measurement, especially around environmental impact and safety in line with new regulations. That’s driving a lot of interest for data-driven applications and software like our own. 

With respect to how things are allocated, we are excited about where software and data are going. However, it is still a relatively small part of the transportation ecosystem today – for the most part, the majority of the funding goes to building more infrastructure towards hardware-based technologies. We see, actually, longer term that there deserves to be a greater allocation and greater interest in investment in software and data companies. So for us, we’re always going to say that there’s not enough money going to data and software, but we see that changing over time. We would love to see it move faster.

Talking about some of the companies who were lost last year and the loss of data there, will we see new companies taking their place?

We’re already seeing that. It’s not so much that all the data suddenly got wiped out – it was just the fact that a couple of leading players were the players that a lot of companies had made a poor decision in building their entire solution around. We took a more diversified approach in terms of partnering with lots of different data vendors, which meant that we weren’t disrupted. There were already a number of substantial data providers that have been in the industry for a long time, and are going to continue to be in the industry for a long time, who are being pretty heavily under-utilized. 

And so a lot of that comes from the education side of the solution providers, of knowing where to look for data. Some of our partners include Michelin, who are backed by Allstate insurance, as well as TomTom, who have been around since 1991 – they’re a massive publicly listed GPS device manufacturer and navigation leader. So we don’t see those companies leaving the industry. It’s just really a question of being good at sourcing, as well as making good business decisions on which partners to work with and which ones to focus on.

We’ve basically seen AI explode over the last few years nearly in every industry. How do you think AI can be used in the transportation industry beyond basically the hype that we’re hearing? 

Yeah, it’s such a buzzword, isn’t it? It seems like everyone today is an AI company. Once you get past the buzzwords, once you get past the hype, the reality is that AI is going to have a major impact on the transportation industry and it will have a bigger impact as long as it’s used correctly. 

What we’ve seen in the transportation space is that AI has struggled, historically. The big reason why is that everyone just expects you to throw an AI model on some data and suddenly you create magic and that just isn’t the way it works. You speak to any data scientist, any machine learning engineer and they’ll tell you the same thing: 90% of our work is data cleaning and data structuring. In transportation,  that’s the problem that’s been plaguing the industry for so long. 

Even though over the last few years we’ve seen an explosion of data, it’s also not been matched by an improvement in the quality of data. So, a large part of what that means is that AI and any automated technology is dependent on the quality, the standard garbage in/garbage out. You can have the best AI model in the world and if you’re putting bad data into it, you’re always going to end up with bad outcomes.

At Flow Labs, one of the things that we’ve really focused on is using multiple types of AI to solve multiple different problems. One of the big ones that we solved a couple of years ago was: How do we use AI to take all of this different data to clean them and to generate accurate data that can build the foundation for other types of AI?

We kind of have three different ways we use AI. We have our AI-driven data integration platform that allows us to combine data from multiple different data sources. It allows us to identify anomalies in data, clean them out and generate an accurate source of truth about what is happening on any given roadway at any given time.

We then have our AI digital twin platform that uses a number of different neural nets, as well as genetic algorithms technologies to leverage that data to generate hyper-accurate simulations of the real world that allow us to create a playground where we can then use other AI forms to test out different scenarios. 

And then we have our third leg, which is our optimization AI. That allows us to test out different strategies, identify the best ones and predict that impact before we can deploy them out into the real world. 

So AI is going to have a massive impact, but it’s really about how we use it. We’ve seen a lot of companies keep riding the next hype wave. Right now, it’s generative AI and LLMs and that was the big talk of 2023. We think they’re going to have a huge impact on a lot of Industries and they will have some impact on transportation, but it’s not the primary problem. 

No agency or transportation manager is thinking “Hey, I wish I could have a conversation with my data.” They’re thinking, “I want to have good insights that can be useful and enable me to make critical decisions,” and generative AI and LLMs really don’t solve that problem. Other types of AI that were around just a couple of years ago are capable of doing that. So it’s really about knowing “what works and what doesn’t?” and also, “which problem and what’s most useful right now?”

Let’s move to another question that might be closer to your heart. So for decades, the data landscape has been dominated by hardware-centric companies and your company is a software and data company. How do you see that changing over the years to come? That question has to do with basically how traffic issues will be more reliant on software than hardware to solve.

Silicon Valley investor Marc Andreessen famously said, “Software is eating the world.” He was talking about the growth of software and how it’s going to impact a variety of different Industries. And we see that transportation will be no different. 

To understand where software plays in this world, you’ve got to really understand the history of our industry. Software solutions weren’t really relevant 20 years ago and rightly so, because there wasn’t really any data for software to really use and software thrives on data. However,  we’ve seen over the last few decades with the advent of “smart cities” increasing amounts of infrastructure that allows software to thrive. There are sensors on every roadway and sensors on every vehicle streaming data. There’s now connectivity that’s better than ever, every single year. It goes from 3G to 4G to 5G so now, we can stream that data from all of these devices to absolutely anywhere. 

And so the stage now is finally set for software solutions to really thrive. And I think one of the things in the future is, it’s going to change pretty quickly because software just simply scales better than hardware or manual-based solutions. Software companies like our own are going from zero revenue to tens of millions in the space of a few short years, which is not something that you can see in the hardware space. That dynamic is going to create a pretty big shift in the ecosystem and it’s going to have a lot of different impacts and changes across the industry. 

But one of the things is, it also depends on how hardware and software can interact. The challenge for players in this industry is how can they best use the infrastructure that is available to deliver exceptional solutions. 

A lot of the time we kind of see this growing talk about a battle between hardware and software. You have one camp that thinks, “The future will have zero hardware – it’s all going to be software.” And you have another camp that says “No, software isn’t going to be doing anything – it’s going to be all hardware.” And neither of those camps is correct, each of those different types of technologies has different strengths that bring a lot to the table. We believe the future is integrated.

So we think that the future is going to require a more collaborative and integrated approach and that’s really what we’re pioneering at Flow Labs, that fully integrated approach that can be the glue between all of these different parts of the ecosystem and return it into something that’s greater than the sum of its parts.

In a related question, one of the biggest topics in the industry is the development of a national digital infrastructure (.pdf) and the concept of cellular-vehicle-to-everything. How’s that going to benefit the transportation systems?

Yeah, so I think when it comes to digital infrastructure. I mean, we all know what the physical infrastructure looks like. We see it every day we drive on it. It’s the tarmac that we drive on all of the different bits of equipment and devices that are out there. Digital infrastructure is in very simple terms, just getting data on what is happening on that physical infrastructure. How many cars are going past this particular road? How fast are they driving? Where are they going from point A to point B? 

That, fundamentally is the digital infrastructure, that awareness of what is happening and the ability to be able to turn that into machine-readable information, so that software and other technologies can leverage that and use it for various different applications. 

Just to give you an example: We had no digital infrastructure 50 years ago. If you wanted to figure out what’s going on in your roadway as a transportation agency, you have to send someone out into the field to visually observe it. They may be able to go down with a clipboard and a clicker counter, but that’s with no digital infrastructure. 

Today we’re capturing data from anywhere up to about 40% of all vehicles at any given time. We can understand their speeds, what their routes are and where they’re going from A to B. We don’t need to go out into the field. We can use the digital infrastructure to capture that information. And now we can immediately put that into the hands of transportation managers and professionals, as well as leverage it in various different AI machine learning models. 

We do think that getting vehicles to communicate with other infrastructure and other vehicles is going to be a worthwhile thing in the long term, but the execution is probably going to look different. At Flow Labs, we’re already able to communicate with close to a hundred million vehicles across the US using the technologies that already exist and the communication stands that already exist. That allows us to open up pathways to actually deliver solutions today rather than having to wait a decade for everyone to get aligned, and for politicians to get funding and allows us to deliver impact today. That’s really kind of the core focus. It should always be about impact, especially when we’re trying to invest political goodwill as well as taxpayer money into initiatives. It should always be about impact and it should be measurable.

Michael Domingo: I want to thank you for your time today. 

Thanks a lot for having me.