Driver workload – the mental and physical demands placed on individuals behind the wheel – is a pivotal factor in road safety. High workload can compromise a driver’s focus and response to unexpected situations, increasing the risk of accidents. Recognising this, researchers at the University of Cambridge are trying a new approach. In collaboration with Jaguar Land Rover (JLR), they have developed a ground-breaking algorithm to continuously and effectively measure driver workload in real-time.
The innovative algorithm relies on a combination of on-road experiments, machine learning techniques, and filtering to precisely assess driver workload based on driving behaviour, road conditions, and individual characteristics. Its adaptability allows it to seamlessly adjust to various driving scenarios, ensuring effectiveness across a wide range of driving environments and driver profiles.
“More and more data is made available to drivers all the time,” says Dr Bashar Ahmad from Cambridge’s Department of Engineering. “However, with increasing levels of driver demand, this can be a major risk factor for road safety.”
The science behind the algorithm
At the core of the algorithm is a modified Peripheral Detection Task (PDT), a method used to collect subjective workload data during driving. Participants in PDT experiments pressed a button whenever a flashing LED light on the car’s dashboard indicated a moment of low workload. Analysis of video footage and button presses helped identify instances of high workload, such as navigating busy intersections or encountering unpredictable driving manoeuvres.
Data collected from PDT experiments formed the foundation for developing supervised machine learning models to profile drivers based on their average workload levels. These models were integrated into an adaptable filtering framework that continuously estimates the driver’s instantaneous workload in real-time, using driving performance signals like steering and braking data for a comprehensive assessment.
Benefits of driver workload monitoring
Accurately measuring and adapting to driver workload holds significant potential for improving road safety and enhancing the driving experience. Understanding a driver’s workload level allows in-vehicle systems to dynamically adjust in real time, prioritising safety and optimising user interactions. For example, infotainment systems could delay non-essential notifications during periods of high workload, allowing drivers to maintain full focus on the road. Advanced driver assistance systems (ADAS) could be tailored to provide more proactive support during challenging driving conditions.
A paradigm shift in human-vehicle interaction
This algorithm marks a significant step toward a more intelligent and adaptive human-vehicle relationship. By continuously monitoring driver workload and adjusting in-vehicle systems accordingly, researchers envision a safer and more seamless driving experience. This innovative approach has the potential to revolutionise how we interact with our cars, paving the way for a future where vehicles become proactive partners in ensuring safety and enhancing journeys.
“This research is vital in understanding the impact of our design from a user perspective,” says Dr Lee Skrypchuk, JLR’s Senior Technical Specialist of Human-Machine Interface. “These findings will help define how we use intelligent scheduling within our vehicles to ensure drivers receive the right notifications at the most appropriate time, allowing for seamless and effortless journeys.”
The algorithm is a new approach that has the potential to significantly improve road safety by reducing the number of accidents caused by distracted driving. It could also be used to improve the user experience by making in-vehicle systems more intuitive and responsive to the driver’s needs, promising a safer and more enjoyable driving experience for all.
The research was published in the journal IEEE Transactions on Intelligent Vehicles.
This article is based on a press release from the University of Cambridge and modified by ChatGPT