DeepTech in map making industry: Eric Bowman, CTO, TomTom

Map making is no longer what it used to be 10 years ago. The advent of new age technologies such as artificial intelligence and machine learning have completely transformed the map making process. With the growing demand for geo-location services across verticals, it has become pertinent to discuss how deep technologies are revolutionizing the industry. Eric Bowman, chief technology officer, TomTom spoke to Dataquest about the latest trends that are transforming the map making industry.

DQ: How has engineering transformed the map making process over the years?

Eric Bowman: I think digital map making was one of the first big data applications. And as such we really got started before all of the kind of data engineering technology which is really quite common place today was fully mature. And it is also a really hard problem. It is much harder to automate many parts of digital map making. And so I think the map making industry has been caught in an interesting position of being both leaders and then eventually by having laterally a little bit followers. And now as we move into the cloud, we have effectively unlimited compute and really a lot of progress that has been made in other industries - from ecommerce to social media - we are able to bring those technologies, and use them to help solve the map making problem.

DQ: Right now there is an advent of deep technologies in almost every vertical. So what is the role that deep technologies are playing when it comes to map making? 

Eric Bowman: On the one hand there are significant opportunities that use really advanced statistics and kind of low fidelity observations. So, at TomTom for example we collect what we call probe data from close to a billion cars in the world.  We are not able to see the individual cars but we can see where cars go and so what emerges from that is a kind of a statistical model of where roads are and how cars are traveling on those roads. And you would think from that that you could just sort of figure out what the road network is but it is much harder than that. Because that signal is really noisy. But now as more and more cars become more and more connected, the fidelity of the data that we start to see is more interesting. And so on the one hand we need to use all of the kind of deep learning, deep data technologies to really pull a really faint signal from this. And then package it in a form that people can really depend on for their safety. 

It is not unlike some applications you know the world of advertising for example, approximate is good enough. Not necessarily approximate is not really ever good enough. It has always got to be converging on reality. And when you really think about the range of human activities that depend on this data, it is a lot more than having fun or buying Christmas presents. It is really getting people from A to B safely and efficiently. 

DQ: How is the demand for geo-location services growing?  

Eric Bowman: We really kind of focus on a few different big sectors. We have had a lot of success in automotive and it has been a tough Corona experience with automotive due to the chip shortage. But it was actually quite a lot of growth opportunity for us because as cars become more and more connected, what we are best at which is really getting people from A to B and more and more with EV, that part of automotive was growing really rapidly. And so EV is really exciting for us and really brings together all of the things that we are best at, and it is a tremendous opportunity. Whereas some percentage of cars would have built in navigations and much smaller percentage would really be meaningfully connected. Pretty much every car in the future is going to be fully connected and fully have automation. That is actually quite an interesting growth area. 

We have recently announced our Indigo product which is a digital in-dash solution. It puts together our technology with all best of breed. Kind of experience that people want. But on the other hand there is so much activity going on in kind of more the big tech and enterprise space. And what we are seeing for example with ride sharing the kind of value creation opportunity that ride sharing companies have seen, through location technology is enormous. When you have literally have tens, hundreds or millions of vehicles, millions of customers and effectively connecting them, and then getting them from A to B, even small improvements of how well we do that ends up being – I mean it is essentially giving millions of more people a better experience. Helping them more conveniently navigate and the value of that is enormous. Also with food delivery taking off pretty much everywhere and lots of entrepreneurial activity, there is such an incredible opportunity for us to help kind of at every level there. And so we also see that as a really big growth opportunity for us. And there are others. But those are a couple of big ones.  

DQ: What are some of the emerging technologies that engineers at TomTom are working on?

Eric Bowman: There is a lot of traditional software engineering. There is a lot of data engineering and there is a lot of data science. And it’s pretty broad data science sort of activities and data science itself covers a number of different fields. But we on the one hand we are working very in like kind of the embedded space and very safety critical applications that run into cars and it should be extremely kind of tight and efficient and compliant against safety standards. ISO 26262 is one of them. And that is kind of one extreme where we work in technology and that is actually really hard. It is kind of like a whole other extreme. We are running pretty large scale spark against the GPS probe data that I mentioned earlier. And we are running lots of parallel route planning applications at scale on trying to understand map quality and really extracting information using deep learning in some cases and other forms of statistical inference and other cases. So it is a remarkably broad range. Much broader range than most tech companies work. And I find that quite exciting. Software engineering in general which I would sort of include with data science is one of those – one of the rare fields where people practicing need to be able to be working across the microscopic nano scale all the way into thousands of terabytes and the dynamic range is kind of enormous. The range of technologies that we use and programming languages is pretty diverse. C++, Java, Python, even some Scala and yeah it is a pretty good playing field technology to choose from. 

DQ: How is TomTom fostering a culture of innovation amongst your engineering talent?  

Eric Bowman: Dev Ops has had sort of an interesting history in the company because historically our map for example and what our customers expected were pretty infrequent. And as we have shifted toward a much more dynamic flow of data and in higher expectations in terms of frequency of updates and freshness of updates as the entire world just expects more freshness, we have had to find a way to apply Dev Ops principles also to data. I think one of the big takeaways from that is as you start to think less about what are the outputs in terms of say target metrics and more about what are the outcomes that we can achieve, one of the key things is to really deliver value as early as possible. And that is one of the great things about even cars moving online where we can start to update software more frequently, customers are taking map updates more quickly. 

We can do something and that effort actually causes something somewhere to change. And the kind of business we were in, in the past, had a lot of built in long cycles whether it was map data or even releasing navigation products. It just doesn’t happen every day. And when you can start to release value all the time, which is what the world has come to expect, it enables innovation in a way that is not totally obvious until you try to go back to the old way and then you start to realize that the real benefit of creating value sooner is that you get feedback from customers sooner and you understand how it interacts. And you get these really tight kind of learning cycles. And those right learning cycles are one of the key ways that we drive innovation. 

You do something and you learn from what you do as quickly as possible and then you do something else and it repeats forever and that kind of creates a kind of a near exponential growth and ability to innovate. And that I think is what we are seeing with the biggest tech companies in the world. They have really mastered how to do that. And we were somehow constrained by the market that we were in until recently from doing that. And so we have really seen that take off. I think the other thing is, how lockdowns and the pandemic have affected how we work and it is much more remote. And on the one hand there is benefit from people being together in the office and being able to work high fidelity and white board situation and it is very productive. 

But on the other hand that introduces these different constraints which office you are in kind of matters. And what we have seen through the pandemic is by working remotely it becomes much easier for people to collaborate around the world more. And that has had a really interesting and positive effect I think on how people really align toward a common north star, working toward shared goals. And then it is not necessarily about faster. It is really about value sooner. So sometimes we do something smaller but we see that value hit and we start to learn from it. And I think we are all kind of adapting how we think and how we work in order to pull those two things together. And it really drives innovation and team work. It is really great to see. 

DQ: Are there any final comments that you would like to add before we finish? 

Eric Bowman: I think we are really quite lucky in the situation that we are in with our team in Pune, India and working with our team in Europe. The time zone differences are a little bit different. A little less than say working from North America. And the opportunity that we see is just so enormous for really solving the world’s location problem and really giving location meaning. And what we find is that – sometimes people leave TomTom and come back because they are just drawn to solving a meaningful problem that really matters for people in the world, and the sense of excitement is so great.