INRIX launches AI-powered traffic reports for TV and radio, replacing humans

In November, INRIX, a Bellevue, Washington, provider of transportation data and analytics, announced the launch of the INRIX AI Traffic Reporter, a platform that “completely automates TV and radio traffic reporting by replacing human reporters with fully AI-generated audio and video broadcasts. “

According to the company, AI Traffic Reporter “delivers instant, accurate updates the moment traffic conditions change, something no human-led workflow can match.” The platform “leverages INRIX’s sophisticated traffic and analytics platform to instantly highlight incidents that impact the safety and congestion of vehicles on the road.”

By leveraging state-of-the-art AI, “natural language and video-game-level graphics technology, customers can now produce traffic reports for TV and radio 24/7, without human reporters or broadcasters, staffing, scriptwriting or manual intervention.”

Powered by more than 36 billion data points per day, AI Traffic Reporter “automatically generates audio traffic reports with synthetic voices, indistinguishable from human reporters, that can be used on radio and TV. The 3D visualizations and synthetic voice fully replace traditional human-delivered broadcasts. By delivering instant alerts the moment an incident occurs, the platform helps reduce secondary crashes, lower congestion, and improve safety.”

The platform translates “raw traffic data into dynamic, engaging formats for drivers, broadcasters, and transportation professionals, ranging from immersive VR displays to mobile and web dashboards.”

In an interview with AJOT, Michael Cottle, Senior Vice President, Enterprise and Automotive Sales, said INRX had provided a mapping/traffic platform app in competition with Google and Apple, but discontinued the service during the COVID pandemic when there were few traffic users on the roads.

INRIX had built up an inventory of data over 21 years, thanks to data provided by mobile phones, road and highway cameras, and data generated by vehicles. That data provided the opportunity to move into the traffic alert news broadcast business.

AI Applications

New AI applications have allowed this data to be aggregated quickly so that slowdowns in traffic conditions can be pinpointed to specific causes, such as an accident or fire, and at the exact location.

Cottle showed a map of the Atlanta area where a typical traffic update could warn motorists about traffic conditions, accidents, and slowdowns using location/map simulations:

“And we have a product called Roadway analytics and Mission Control that allows you to kind of look back at the history and see what the traffic conditions were on a particular road segment that may have caused disruptions to your schedule.”

At the same time, he said: “There's some dynamics happening in the broadcast industry that … with the user base migrating to some extent to streaming services and to other apps, the ability to monetize traffic has become more challenging. The population of users is decreasing slightly, but the costs are staying the same for producing that content: You've got to collect it, you've got to create the scripts, then you have to broadcast and put it on air and we've been able to automate.”

This then is fitted into standard radio formats, “So, it's now just completely push button, and you can have an experience for the listener that's just like a DJ reading it out over the air, mixed with bumps that have ads or a bed that's maybe a music underlay. So, this is now in production with a number of radio stations. And it's either helping them reduce their cost of production or relieving the DJ from having to do this work themselves. While essentially an AI voice is reading out that bulletin as they would.”

The product can also be used for television: “We also have visuals for TV.”

So, the ability to “essentially produce a map or display a map that zooms in with a voiceover that shows where these incidents are taking place, where there's congestion, where there's road construction, etc. The AI broadcasters also have an understanding of severity. So, it understands where, let's say, a region like Atlanta, where the most impactful incidents are taking place, and can essentially sort through those to bring up in that 30-second spot. Zoom in the map, show that location, and animate. If it's, let's say, a construction situation, show construction crews on the street. If it's an accident, show cars on the street. Completely created from gen AI technology.”

Production was “developed in-house with the ability of technologies to create 3D maps or 2D maps; it's very easy to do flyovers. So, you push a button, it sorts through the incidents that we collect naturally as part of our traffic feed, creates the script, inputs that into a text-to-speech engine that uses a natural voice, either of the talent that usually does it or some other localized voice, and then coordinates the audio with the flyover to the various locations.”

Future US Market

INRIX currently does not have any customers in the United States because it is still testing the product in the UK. “We have no customers in the US right now for either the radio or TV solution, just because we just launched it, and the UK was our test bed to really work out the kinks and get some feedback from the radio stations.”

INRIX combines several data sources: “So, we receive several signals that enable us to detect an incident on the road. Those sometimes come in through 511 in the United States. And then the other way, obviously, is we can see that signal come in through the probe data from the vehicles. Probe data refers to the data that's being sent back up from the cars. Or from mobile applications that show how fast you're moving on the road. We're getting that in real time with very low latency, often just less than a few seconds of latency between when the car experiences it and when we receive it, and so as we start to see those red areas on the map show up, we can then also investigate that.”

An INRIX team supervises the data analysis: “So, we have a team of over 100 people in our operations center that analyzes those situations. They receive a signal, and they investigate. Essentially, it's called the Index Investigation group that goes about looking at maps, looking at Twitter feeds, looking at cameras from the Departments of Transportation, and validating what's happening on that incident. So, all that gets created sometimes with human intervention, sometimes automatically to feed either the traffic that goes back into your navigation system or for these on-air broadcasts.”

There are ways to fake traffic situations: “There have been a lot of cases where companies or individuals have essentially faked the system, faked Google by indicating that there's an incident or closure that really isn't there. They can put 10 phones in a car and drive slowly along the street. Sends a lot of signal back to Google that, oh, there's a slowdown here. So, you can manipulate that situation.”

Media organizations are responding positively to the INRIX traffic broadcast product:
“They're seeing the pressure they have from a cost basis to improve the efficiency and the cost basis for delivering traffic bulletins. They're also seeing the productivity gain that it can provide for their talent, so freeing them up to do other things. Since its introduction less than a month ago, we have had really positive interest from customers in Asia, Australia, Canada, the U.S., and as I said, we've already implemented in the U.K...”

Cottle said there are “I believe 130 stations currently in the U.K.” using the INRIX traffic broadcast product.

INRIX was founded in 2004 and says that it “pioneered intelligent mobility solutions by transforming big data from connected devices and vehicles into mobility insights.”