Modern cars and connected smart systems have a lot of information that they can pass on to the driver, from traffic alerts and directions to incoming calls and messages.
It is not always safe or appropriate to do so, however, especially at times when the driver particularly needs to concentrate on the traffic or conditions in front of them.
Now, researchers have developed a system that is designed to continuously monitor driver ‘workload’, allowing decisions to be made in real-time on when it is safe for information systems to interact with the driver.
The Cambridge University researchers, working in conjunction with Jaguar Land Rover (JLR), used a combination of machine learning, experiments conducted on the road, and a technique known as Bayesian filtering, which involves making predictions based on incomplete or uncertain information.
Using this combination, they were able to develop an algorithm that they describe as highly adaptable and able to respond in close to real-time to a number of different factors.
These include driving conditions and road type, changes in the driver’s behaviour and status, and individual driver characteristics.
Algorithm can feed into in-vehicle communication systems
This information on the driver’s current mental workload can be used to inform in-vehicle systems, including infotainment systems, navigation, displays and advanced driver assistance systems (ADAS).
The delivery of any notifications or interactions can then be customised with safety in mind.
Drivers may only be alerted when the workload is comparatively low, for example.
While algorithms have previously been developed using technology such as eye gaze trackers and biometric data from heart rate monitors, the researchers wanted to use information that’s available in any car.
This involved monitoring driving data related to braking, acceleration and steering, as well as data from participants who were asked to push a button worn on the finger at certain times to indicate low-workload situations.
The collected data was then combined with machine learning and Bayesian filtering techniques to produce the algorithm.
Dr Bashar Ahmad, co-first author of the study, which is published in the journal IEEE Transactions on Intelligent Vehicles, said that more and more data is being made available to drivers all the time.
With increasing levels of driver demand, however, he added that this can be “a major risk factor for road safety”.
Today’s news was brought to you by TD SYNNEX – the UK’s number one solutions distributor.
