Using Artificial Intelligence to Improve Road and Driver Safety

Millions are affected worldwide by reckless driving and speeding. The World Health Organisation estimates (using data collected from 180 countries) that 1.25 million people die from car accidents worldwide each year. Sadly, many of those accidents were likely preventable.

Thankfully we now have technologies that can help significantly reduce those numbers, by alerting drivers to the dangers around them, their own behaviour, and even taking control of some aspects of the car in order to prevent an accident, with things such as lane assist or cruise control for maintaining speed and safe driving distance between cars.

By using Artificial Intelligence (AI), companies like Hyundai, Nissan, Renault and Tesla are creating ‘connected’ vehicles, which can provide live (real-time) navigation to drivers, geographical risk warnings (like sharp corner ahead or a road block due to flooding or a land slip), monitor speed and provide speed limit information to drivers, and analyse driver activity including fatigue detection.

This provides feedback to suggest training to improve driving skills in targeted areas of need. These features will help people to drive in a safer way with more awareness of their surroundings, which in turn make the roads safer for everyone.

The information gathered could also provide evidence for insurance companies regarding the skill of drivers, potentially rewarding safe drivers with lower premiums in the long term. Using AI for safety in vehicles is a fairly new, yet fast growing industry, which has made significant leaps forward over the last few years and is predicted to become mainstream and common use in the very near future.


A good example of this technology is ‘Driveri’ (founded in the US), currently working with car manufacturers mostly in Asia. The company CEO and co-founder is Avneesh Agrawal, who created the Driveri platform through his company Netradyne.

It was a start up which raised over 21 million USD in the initial fundraising campaign. In late 2018 Netradyne formed a partnership with Hyundai MNSOFT to develop HD GPS maps for their future vehicle models and other uses. They are currently working together to map out the whole of the United States and are making great progress with the project, using their own and crowdsourced data to create the type of High Definition maps that will be needed soon for self-driving cars to function properly. Roadshow has written an article about it here.

The AI technology used in the Driveri system continuously records and analyzes drivers, reporting back to commercial fleet managers, providing them with extensive data on things such as the distance drivers keep between other vehicles, speed, staying in lanes and more.

It works by integrating hardware like internal and external car cameras, with GPS data, speedometer data, and an onboard computer device with advanced software, all linked to the cloud, and gathering data and uploading data live for the driver, and the company motoring their driving. Feedback can then be used to identify areas of need for driver training, reward safe drivers and set goals for the future.

Technology similar to Driveri is currently being developed by Nexyad (based in Europe). The Founder of Nexyad claims their software can predict 75% of all accidents, send drivers an alert, give them time to slow down and reduce accidents by at least 25%, with that percentage increasing as they add more features to their software and devices.

Due to computation power increasing rapidly and the cost of internet connectivity decreasing, these types of technologies are sure to rise-up and become more common and effective.

Over the next few years we can hope to see road toll statistics drop in countries where cars with this technology are available.

With companies like Tesla creating self-driving cars, and constantly evolving AI technologies like Driveri and Nexyad, GPS devices and apps that have live updates becoming more common, the roads should soon be a safer place for everyone.

For Further Reading: