Predicting and preventing road crashes with AI

A collaborative project in Sweden is using AI to predict when collisions on the road might happen and therefore prevent them.

The connected traffic system gathers a huge amount of data. The premise of the AI Aware project was simple: how could all that data be used to look beyond real-time and predict accident risk levels?

Connected vehicles, sensors, mobile data and other sources can tell us a lot about what is happening on the roads. The potential for collaboration between traffic authorities, traffic management providers, map and location data providers and OEMs to make roads safer is huge. 

Predictive awareness through the power of AI could become as important as the safety belt for road safety.

AI Aware, a joint effort between Volvo Car Corporation, HERE, Carmenta Automotive, Ericsson, Zenseact, Trafikverket and the Gothenburg City and Swedish Transport Association, uses this information to notify connected systems ahead of time when there is an increased risk factor.

“This takes us one step closer to the vision of zero accidents in traffic," said Johan Amoruso-Wennerby, Business Strategist at Volvo Car Corporation. “It could become as important as the seat belt."

AI Aware uses predictive AI algorithms to detect when there is a risk of a vehicle colliding with another vehicle, a pedestrian or other obstacle. It then sends predictive alerts to vehicles to prevent the crash from happening.

Petter Djerf, Senior Account Executive at HERE, explained: “One reason we can do this now is that we have a bigger amount of available data than in previous years."

The project has focused mainly on two kinds of data. One was dynamic data, including traffic data, weather data and other rich data sources that are changeable and temporary in nature. The other has been maps with higher granularity than would have been possible a few years ago. And the key to its success was the way the organizations involved in AI Aware shared data with each other.

Drivers receive alerts warning of bad weather conditions, obstacles ahead or other dangers.

Kristian Jaldemark, Vice President, Head of Development at Carmenta, said: "It is wonderful to have this ecosystem and the data available to create these services. The AI Aware project has really shown the potential of sharing data between companies to create connected services for increased road safety."
 

When combined with the power of the cloud platform, and the ability of software inside vehicles that can handle information and take decisions based on it, the data can be put to work to prevent crashes before they happen.

Read more: Sweden's one-minute city success story

Beyond the line of sight

AI Aware has moved beyond the proof-of-concept stage and is now operational in Gothenburg, Sweden. “Everything we have done is in place and it works," Petter said. “There are agreements between the companies involved and we comply with GDPR."

Now, the task will be to scale the project and add new datasets where needed. AI Aware could then be used throughout Sweden and beyond.

It is not just individuals who can benefit from these alerts. Because the project can look for patterns in the data and predict risk, it could be applied by city authorities to improve the traffic system and make mobility schemes function better, or by commercial fleets looking to improve safety and efficiency. Technology from the project could be used to decide which roads and when autonomous or highly automated driving is suitable or possible.

“This can also reduce the carbon footprint with smoother traffic and improved mobility," said Johan from Volvo Car Corporation.

The AI Aware project is the latest in a series of five related projects for Drive Sweden, a strategic innovation project that looks at ways to increase sustainable mobility.

Enhance the driving experience with real-time information. Help drivers to arrive safely, comfortably and on schedule, by using dynamic content APIs in your apps and services.