October 6, 2021
Your data is wrong.


This post is Part 1 of our series on the critical advantages of data integration.
"
Everyone in traffic management wants to be able to predict the future, but today, nobody can even measure the past or present."
Jatish Patel
CEO
Flow Labs
Your agency has spent millions of dollars on state-of-the art IOT sensors, top of the line software and cutting edge AI and automated solutions, and they’ve often left you feeling like this.

Not only do these solutions not solve the problem you wanted them to solve, they create more problems. Yet another piece of equipment to manage and maintain, and the anxiety of yet another thing that could go wrong every day. There’s a fatal reason why most automated solutions haven’t worked: your SPMs, your adaptive signals, your simulated models. Why you spend so much time in the field, why it takes so long to diagnose problems, why you’re getting complaints about problems you didn’t know you had.

The data you’re collecting from hardware is inaccurate and unreliable.

Over the last few decades, transportation agencies have invested billions of dollars on deploying state-of-the-art detection devices and IoT sensors on roadways. Agencies rely heavily on the data generated by this hardware to make mission critical decisions. Similarly, your artificial intelligence and automated solutions are using this data to measure and optimize your roadways. 

In partnership with Utah Department of Transportation (UDOT), Flow Labs conducted a study of 435 roadside detection units. While manufacturers typically claim a 90+% accuracy rating, we found that only 27.6% actually met this threshold. Most concerning of all, agencies have no way to know which detectors are counting correctly and which are malfunctioning: no way to know which data to trust and which data not to trust. When only one out of four detectors produce data reliable enough for field conditions, how can any traffic professional have the information they need to improve their roadways?

With inaccurate data, road users, traffic professionals, and agencies suffer.

When your data is inaccurate, your agency is reactive, instead of proactive. Since your data isn’t actually showing where your roadways are congested and unsafe, you don’t know where your biggest problems are. Instead of being able to identify these conditions immediately and addressing them before they impact road users, you’re only aware of these pain points after they cause an issue, and you’re relying on citizen complaints and time-consuming fieldwork. 

When your data is inaccurate, it’s impossible to track your project’s ROI and to manage traffic at scale. You have no way of knowing whether the changes you made worked, and, if they did, why. 

When your data is inaccurate, your AI and automated solutions won’t work. Even a perfect model won’t produce usable results if the data input into it isn’t usable. 

When your data is inaccurate, your road users will continue to encounter congested and unsafe roadways, causing massive liabilities for agencies. Preventable crashes, fatalities, delays, emissions, and inequities will keep occurring.

Data integration is the solution.

You shouldn’t need to buy hundreds of more detectors or invest in an entirely new type of detection. These approaches are costly, inefficient, and ignore a core truth: there isn’t one dataset that can tell you everything you need to know about your roadways. 

Instead, the solution lies in a platform that can integrate any dataset together, and create artificial intelligence to produce data that can be more accurate, more complete, and more relevant to agencies than any single source. Today, while still relying primarily on hardware, agencies often do collect data from a variety of sources, such as information from data-rich Connected Vehicles (CVs). However, all of this data is siloed from one another, as agencies use different datasets to solve different problems: for example, you might use hardware to manage your signal timings, but have a CV initiative to improve vehicle safety. 

At Flow Labs, we’ve proven our AI algorithms can successfully clean up datasets. Our Virtual Sensor technology seamlessly integrates sensor, signal, and CV data to produce data that’s 94.4% accurate in field conditions. Second, as a hardware free platform, we’re accomplishing this right now. We’ve proven agencies can access high fidelity data today, using the existing hardware and infrastructure you already have. 

Our series on data integration will first discuss the accuracy of our Virtual Sensor technology in more detail. Then, we’ll show you how we’re combining our highly accurate data with more artificial intelligence to solve the most complex problems in traffic management. We’ve developed industry leading analytics, which provide deep insights into safety, operations, environmental impact, and mobility. We’ve developed a breakthrough proactive monitoring platform, which can alert agencies to poor conditions and malfunctioning equipment before these issues become problems. Finally, we’ve developed accurate, on-demand signal optimizations that are multimodal, allowing agencies to optimize for both safety and mobility. 

Because it isn’t enough to provide agencies with the most accurate data available. This data must be presented clearly, and in a way that empowers traffic professionals to take immediate action. Furthermore, this data must be used to solve problems in traffic management instead of only showing where the problems are. 

And we’re just getting started.
To see our virtual sensor technology in action, please contact us to schedule a demo at sales@flowlabs.ai.
Thank you!
Your submission has been received!
Oops! Something went wrong while submitting the form. Please try again.