Scaling up Traffic Safety and Awareness using Data-Driven AI

Scaling up Traffic Safety and Awareness using Data-Driven AI

The innovative project ‘AI Aware - Scale Up’ has shown that we can predict dangers on the road using live data.

Approximately 44,450 motor-vehicle deaths occurred in 2023, according to the National Safety Council in the United States. When we consider the larger picture of life-changing incidents while on the road, the total number of fatalities and serious injuries goes way beyond the initial data. Our roadways must become safer to protect the lives and health of the public, communities, and drivers alike.

To confront the challenges of road safety, the shared concept of Vision Zero embodies the principle of completely eliminating road-related fatalities or serious injuries while using road networks. Vision Zero has been adopted in different countries and fuels many of today’s strategies for making roads safer, healthier, and providing more equitable mobility to all.

In November last year, we saw the conclusion of the year-long project funded by Vinnova via Drive Sweden and Future Mobility known as AI AWARE – Scale Up that aimed to better understand roads and traffic dynamics by putting together data from numerous sources to provide a way of predicting events in a traffic system.

The AI Aware - Scale Up video presentation of the project’s aims and potential. Source: Drive Sweden

The project was one of the first of its kind to challenge and go beyond the system boundaries to influence how road safety is approached. The initial focus of the project concentrated on the Swedish perspective of the problem before being expanded to California to better understand how the project adapted to different contexts. Different countries have different roads, and ultimately different driving behaviors.

Historically, the approach to road safety started with the car as a single unit as part of a larger system. Technological limitations prevented experts from having a more complex understanding of road networks and traffic. However, the AI AWARE - Scale Up project presented a way in which improved vehicle interconnectivity can improve both system performance and the traffic system as a whole.

The results of the project were presented to a broad audience of experts and a recorded version is openly accessible to watch on Future Mobility’s website. In this article, Sengerio explores the evolution of the AI Aware project and the potential applications of this new cutting-edge technology that will shape traffic management and road safety, as well as the challenges that lie ahead.

Big city skyline at night

AI Aware – Scale Up: The Road to Vision Zero

The AI Aware - Scale Up project aims to create safer roadways by developing solutions for predictive awareness that can contribute to avoiding dangers and accidents in traffic.

The project was spearheaded by Drive Sweden and other partnerships to conduct each area of the project. These include Volvo Cars, Carmenta Automotive, HERE Technologies, Zenseact, RISE, and the Swedish Transport Administration.

After the promising results from the earlier phases of the project starting in 2017, which tested the predictive abilities of a collaborative cloud environment and the efficiency of central traffic control in supporting connected and automated vehicles, the project was expanded beyond its Swedish origins to California where new use cases were added to contribute to increased traffic safety.

During the project’s final conference, Johan Amoruso Wennerby, Business Strategist at Volvo Cars, reiterated the project’s main objective to render the traffic system a safer place.

“We have really been focusing on the accident and risk estimation. Safety has been the heart of the project and we believe that there is the possibility to understand accident risk and risk level in a dynamic way for all road segments. We believe that, by analyzing and utilizing data in a good way, we must be able to understand what is the actual risk level and how will the risk level in the near future be for a certain road segment. This is important for manual driving, for autonomous driving, and also relevant for other stakeholders in the mobility and traffic system.”

Futuristic image of an AI driveway

Sharpening Predictive Traffic Safety

When it comes to predicting a fatal accident, simply observing the total number of previous fatal accidents in a given scenario provides only a somewhat limited perspective of the problem. That is, it fails to capture the complexity of the interdependent factors that make an accident more likely to occur at a given time.

In order to understand the more viable probability of an accident happening, the project implemented a model that collected and centralized different data streams from its partners to have a broader understanding of the conditions of the environment and traffic system that may affect driving behavior and roadway safety.

The different data streams paint a more precise picture of the roadway system and driving behavior to allow risk analyses and incident prediction. With this model, researchers can calculate the risk of an accident within a 100-meter proximity as well as the risk of an accident along a specific route. As a result, drivers would also benefit from receiving this information to adjust their driving style or route. For instance, if a road has an increased risk of an accident due to external weather conditions, a driver can decide to take a safer alternative or adjust their speed accordingly.

Symbolic view of ai in traffic

The use of multiple data streams raises the question as to how the data can be interpreted and operationalized so that researchers can arrive at conclusions quickly and efficiently. The project utilized Carmenta Automotive’s TrafficWatch, a cloud-based system that provides safety and situational awareness to connected fleets.

The TrafficWatch system was deployed inside Volvo’s cloud and controlled the data that was coming in and out from each connected vehicle. This is because it was deemed a safer solution and allows researchers to examine additional data from Volvo cars and combine this with the third-party data. TrafficWatch then combined the data obtained from Volvo cars with the third-party data, provided by HERE, to lay out a centralized platform where all the different datasets can be observed.

The platform, which was presented during the final conference and observable here, offers a simplification of the different data streams, allowing researchers to easily interpret real time roadway information from one place. This information can then be filtered or segmented to identify specific variables, such as accident risk situations (i.e. Slippery roads), hazard light alert situations (when a car flashes its hazard lights) or the whereabouts of emergency vehicles, as opposed to viewing all the information at one time.

Exploring Data Governance and Policy Challenges Across Continents

Data sharing and artificial intelligence are having a significant impact on policy across various domains. Advancing technologies raise complex ethical and regulatory questions that policymakers have to grapple with in order to develop regulations that ensure they are used responsibly and ethically.

Legislation, or the lack of it, can affect the development of AI and access to data. Therefore, establishing regulatory frameworks can assist in maximizing the benefits of new technology while minimizing the risks, protect fundamental rights, and transform ethical principles into law.

By putting together a virtual workshop composed of both internal and external participants from the U.S. and Europe, experts could discuss policy and regulatory issues regarding vehicle-related AI and data access. This included defining what artificial intelligence is and how it should be regulated. Moreover, they compared the rules and policies in Europe to those of the U.S. given that each area utilizes a different law system (i.e. Europe’s civil law system versus the common/case study law system of the U.S.).

Vision of an AI-driven car
Source: Drive sweden

For the project to be adapted to the Californian system, there was a set of important data types that Europe accessed that also needed to be available in the U.S. for the same purpose. These included data types concerning infrastructure, that is, the road network links and their physical attributes; data on the state of the roadway networks to understand weather conditions; data types on the real-time use of the network, such as traffic volume, speed, and travel times; and finally, access to road safety-related events such as temporary slippery roads, obstacles in the road, and short-term road works.

One of the main conclusions from the project’s workshop was that there are still challenges ahead for sharing data between different entities and creating stronger policies. While in both the U.S. and Europe there are different approaches to accessing data, data sharing is often based on either legal requirements or economic incentives, not solely for the benefit of society.


There’s a shared responsibility on the roads and, in order to achieve a true Vision Zero system, different players need to be responsible for different things. While the project has demonstrated how data sharing helps to make the roads safer, it also unlocks a wider scope of systems performance and traffic system as a whole. AI Aware - Scale Up steers away from the paradigm that views the single car as the central part to the narrative of improving road safety but rather as a part of a larger, more complex system in order to add another dimension of vehicle interconnectivity and explore the potential value this can bring to predictive road safety.

The project has put forward a model that offers an analytical lens that changes the perspective of how road safety is viewed, while at the same time limiting its true potential to the data only being provided by Volvo’s fleet of vehicles. In addition to crossing continents, the project and its model would be greatly supported if other vehicle manufacturers get on board and share Volvo’s perspective on the future of safety.

With more manufacturers willing to share their data, this would amplify the understanding of road risks and dangers while, at the same time, testing the flexibility of the project’s model to understand whether it would maintain its current cloud-based model or if it would eventually turn to an interconnected cloud-based system between different manufacturers. Moreover, it will be interesting to see how policymakers can incentivize manufacturers to share their data while protecting the overarching well-being of society.

Lastly, what would happen if the scope of this technology were increased? The potential of interconnected traffic could drastically sharpen roadway safety and the mobility ecosystem as a whole. Perhaps we can begin to imagine safe speed warnings due to slippery roads that could turn into automated cruise control, or automated vehicles that could select a route to minimize the traffic impact and make safer alternatives. Only time will tell what this technology holds — but it sure looks exciting.

Scott Frankland

Scott Frankland is Head of Content at Sengerio. His spirit of inquiry leads him to the world of transportation and mobility to connect with the industry’s leading experts and shine a light on the hot topics.