Future data: new technologies to solve ever more complex problems

In a time of rapid change in the transport sector, Paul O'Neill, Director, Intelligent Data, considers the opportunities for improving data collection, processing and compliance

Things are changing fast in the transport sector. The rate of change we're seeing now is probably more rapid than I've experienced in my 20-plus years in the industry: lifestyles are changing with advances in digital technology, and Autonomous vehicles and Mobility as a Service (MaaS) likely to change the way that people move. So Intelligent Data are engaging more and more with new technologies to create solutions to even more complex problems; recently we have added drones to the CCTV, ATC and ANPR equipment we already use. 

Intelligent Data will be exhibiting at Smarter Tomorrow

Intelligent Data was a pioneer of real-time data collection and aims to remain a tech leader in the sub-sector. We were the one of the first players to invest in and integrate equipment in a significant way, using Home Office-compliant cameras with a high level of accuracy; capturing more than 90 per cent of number plates, when other were using laborious manual processes. With major computing power and fast mobile networks coming on board, our core team will be able to collect and process much larger, more robust datasets than we ever have done. 

I can see products coming online that will change the way that data is collected; equipment that is not available to the market today. It might be sensor-based technology, it might be a hybrid technology using Artificial Intelligence: Artificial Intelligence will be a game changer. 

We are, and will remain, at the forefront of investing heavily in technology to make sure we remain pioneers, either by developing products and software or by integrating existing services. Our mission will remain the same, to collect the highest quality data in the safest manner possible, backed up with excellent customer service. 

Our clients will benefit from bigger and better datasets thanks to the investment we have made in our software systems. We've spent significant sums of money interfacing our products with the technologies that our customers are using. So our real-time ANPR products, for example, interface with some of the leading software packages used in this sector. We have our own software development team creating products that not only improve quality, but also visualise and enable quicker access to datasets. We've developed software around all the steps in the processes that our customers would normally follow. We collect data, and our software outputs it in a proprietary software format. So, for example, from a 10 million vehicle ANPR survey, we can visualise prior simulation model matrices in real time.

Timescales are also shrinking rapidly, thanks to technology. For example, a few years ago we collected a significant amount of data for a project – around 22 million ANPR records –which would have normally taken about six months to process. Yet we were able to speed up the time taken from data collection to data delivery significantly, delivering processed data in a couple of weeks. We are looking at supplying datasets that might be 100 times the size, and significantly more robust and more accurate, than those that modellers had access to even five years ago.

Data value

Data is valuable. I don't have any issues with open data, but there must be a level of quality assessment before it's passed to third parties. I know that our data is very accurate because we subject it to rigorous QA processes, and that adds real value. There has to be a level of robustness in the data to make sure that it's fit for purpose. I'm all for open data, and for sharing data, but it's important to know that data suppliers are very different in terms of their QA processes. I know the quality and rigorous processes our data goes through before we pass it to the customer.

Open data such as that provided by TfL can be really, really powerful. I've seen an example recently using some rail data, looking at where the congestion points are on the rail network.  Open data is useful in the identification of issues: I was reading the other day that an Amazon executive was about to consign a product to history based on gut feeling, but his team analysed the data supporting it, decided it was of value and launched it, and a year later it added $2.3 billion to their revenue. Data is there to support better decision-making, and we need to place a high value on this. It's great to focus the mind at a kind of pre-feasibility level to look at where investment might be best placed. But in terms of the granularity of the data and digging into detail, even at pre-feasibility or feasibility stage, I'm not so sure that much open data can be used to that level of quality. I founded Intelligent Data 12 years ago, when working as a modeller for a large consultancy, simply because of the quality of data I was receiving, the practices employed to collect the data, and levels of service were really poor. A lot has changed in the sector in that time. The future is positive: the quality of data capture, processing and preparation is key to transport modelling, and new technologies are likely to improve all three functions in future. 

Intelligent Data has carried out a significant amount of low emissions zone studies across the UK. Owning a significant amount of kit, we deploy real-time digital ANPR, up to 200 - 300 cameras at a time over a 4G network, which sends the data straight back to our server. We can identify the percentage of low emission vehicles on the road, the number of electric vehicles; even how many BMW 3 Series are using a given piece of road on a given day, by the second. We work at that level of granularity.

Working with drones

There are major issues at a local level and on major arterial routes, and drones are proving to be very useful for clarifying the causes. With a queue of 3.5 kilometres long, it's virtually impossible to use cameras to look along the length of queue and see how drivers are behaving. So we deploy drones with 4k imagery capability and image stabilisation, for 12 hours or so, to film from 400 feet, and we can pick up an incredible amount of detail. Drone data really helps decision-makers understand what's actually happening; in many cases, once they've looked at the data, they see that their hypotheses need refinement. Using a drone, we can actually see what is causing the issues. We have deployed drones at roundabouts, for example, to study supposed exit-blocking, but seen that exit blocking wasn't an issue at all, it was actually all related to signal timing. At this level of detail, drones offer real potential, at the pre-feasibility or feasibility stage of a study, to both supply great data for analysis and offer great value for money.