New research opens door to fleets using AI accident prevention, says FleetCheck

New research means real time accident prevention made possible by artificial intelligence (AI) could become practical for fleets in the next few years, FleetCheck is predicting.

While AI based collision prediction has existed in research for a number of years, adoption has been hindered by two major challenges, explained Neil Avent, chief technology officer at the fleet software specialist.

He said, ‘First, many systems cannot explain why a particular situation is considered risky. Second, the computing power required can make them impractical for real time use. 

‘What makes the new model – described in sciopen.com – interesting is that it can explain the factors behind its prediction while also reducing processing delays. In our view, this opens the door to fleet adoption. The technology is now fast enough to be practical and also reflects how fleets approach risk management. 

‘Compliance means that it is not enough for the AI fitted to a vehicle to flag a warning, it needs to be able to explain why in a manner that is understandable and accessible to the vehicle operator. That issue now appears to have been resolved.’

The model is said to be able to predict around 85% of all potential collisions, a figure that Neil said would catch the attention of fleet managers.

‘The potential ability to prevent 17 out of 20 accidents is something that fleets cannot ignore in a risk management sense. That is why we believe car, van and truck operators will be keen to adopt the technology if it proves effective in real world conditions.

‘One area that needs consideration is what kind of information operators might need in terms of the AI decision making process from a compliance point of view, something which could require a wider debate across the industry. Some basic points could include the factors behind a situation being considered risky by the technology, including the on road conditions, other road users, weather, the attention of the driver, current speed and more.’