Synthetic Data: A Solution for Improving Driver Safety

Road safety continues to be a top priority, and it’s no wonder why. The numbers are dire: in the U.S. alone, it’s estimated that in 2020, over 35,000 fatalities occurred as a result of motor vehicle traffic crashes. In addition, more than 3,000 fatal crashes were caused by distracted driving in 2019. For many of us, driving to work, dropping the kids off at school, or running to the grocery store becomes another routine on a normal day. So, it’s easy to shift our focus to the ever-growing distractions of modern life or try to multitask on the road. The problem is, we tend to forget just how big of an issue these little distractions create. 

The European Union (EU) has made distracted driving a top priority.

Beginning in 2022, all new cars entering the EU market must be equipped with advanced safety systems. Among the mandatory safety measures is distraction recognition and alert systems on trucks and buses to warn when vulnerable road users, such as pedestrians or cyclists, are in close proximity. The European Commission expects that the proposed measures will help save over 25,000 lives and avoid at least 140,000 serious injuries by 2038. However, meeting the new safety regulations will not be an easy task for car manufacturers. Carmakers will be faced with dedicating vast amounts of resources to build and deploy cars to collect diverse datasets to train AI models. A dizzying array of environmental conditions and edge cases must be engineered to be robust enough to reflect the real world, from geographical features and landmarks to living beings. Besides being an expensive and time-consuming process, it is difficult – if not impossible – to obtain sufficient examples of diverse sets of drivers across a wide variety of situations. This is a significant challenge and barrier to using real-world data to power these AI models.

Luckily, AI can come to its own rescue with synthetic datasets – computer-generated simulations that ensure an ample supply of diverse and anonymous training data. Manufacturers will be able to mimic driver behavior in virtual car environments to test and iterate their models across a broader set of settings and situations without having to build and deploy fleets of vehicles. Synthetic data enables the development of necessary data assets at a fraction of the time and cost of externally acquiring hand-labeled training data. Not to mention, it significantly reduces the difficulties associated with the preparation, labeling, and collection of data for AI applications in cases related to driver safety.

The emerging driver safety technology is already becoming an essential component of autonomous driving and computer vision AI systems. By bringing together techniques from the movie and gaming industries (simulation, CGI) together with generative neural networks (GANs, VAEs), car manufacturers are able to engineer perfectly labeled, realistic datasets and simulated environments at scale. Since the data is generated, the image data comes with an expanded set of pixel-perfect labels, including facial landmarks, gaze, angle, depth maps, segmentation, surface normals, and facial meshes. As a result, automotive manufacturers will be able to build more robust training models in a fraction of the time and cost of traditional human-annotated real-world data approaches.

Each wave of AI innovation builds upon the last, and the opportunity for synthetic data in the automotive industry is no different. High-quality synthetic data for computer vision systems can help advance security systems for car manufacturers and reduce road traffic accidents not only in the EU, but all over the world.

[To share your insights with us, please write to sghosh@martechseries.com]