There is a huge drive by governments, regulatory bodies, academia and the automotive industry to introduce new measures to improve vehicle and road user safety. The case is obvious – the number of global road traffic deaths is estimated at 1.35 million each year with a fatality every 24 seconds. Focusing on preventable incidents – such as those caused by distraction and drowsiness – reveals the need to consider new approaches.
A driver monitoring system (DMS), or occupant status monitoring, is one of the new safety measures that will be introduced into vehicles as an active safety measure. It serves as a front-line ADAS that leverages cameras or other sensors to monitor driver behavior and assist in driver attention management. It is a recommended strategy to managing the risks associated with distracted and drowsy driving in non-automated vehicles, a future requirement for automated vehicles, and is a necessary solution to meet international goals for reducing road trauma.
Driver monitoring technologies standards
Monitoring the state of the driver is a primary safety measure for all forms of vehicle use. Driver monitoring can potentially be done using different sensor inputs, such as steering behavior using a steering wheel torque sensor. However, there is wide consensus that camera-based driver monitoring is the best approach.
Driver monitoring-based technology, and support for it, is not a new concept brought on by the automated vehicle craze. In fact, usage of driver monitoring technology dates back to 2007 when Toyota and Lexus deployed systems in consumer vehicles to aid collision avoidance. More recently, governmental support has grown. Euro NCAP has, in its 2025 Vision Zero roadmap, specified use of safety features including DMS. This approach has been further substantiated by European Parliament direction, which has called for mandatory installation of DMS, along with automatic emergency braking and lane keeping assistant.
Three considerations for driver monitoring technologies
Clearly, there is a recognized industry need for driver monitoring to support safer road use in both automated and traditional non-automated driving modes. This need is evident through several EU-level proposals that call for the mandating of this technology.
There are, however, many questions and issues that need to be addressed to help move the industry from policy to practice:
1. What should the minimum viable technology concepts be? Drowsy and distracted driver states can be defined and measured in different ways – how they’re defined and operationalized, and the implications of different approaches, is a critical first question that needs to be addressed.
2. A related issue lies with the Human Machine Interface (HMI), or simply, how and when should you inform the driver of their impaired state to achieve the desired change in behavior? This is particularly the case for drowsiness, where our own research has shown that a number of levels of feedback to the driver are required to maximally reduce the rate of drowsy driving events. A key issue impacting system effectiveness is around measurement thresholds and parameters, and systems with low specificity may have high false positive rates.
3. What should the testing protocols be for this technology? Driver monitoring is a technology that monitors the state of the driver with the intention of managing driver attention. We have seen from our experience in automotive OEM production programs that a fundamentally new and different approach to technology validation is required. Key issues to consider here include the test scenarios, driver characteristics, how to elicit driver states such as distraction, drowsiness and engagement in a testing protocol, what methods to use to measure them, and, of course what the performance standards should be.
The future of automotive safety
Many regions in the world are striving to implement variants of the Vision Zero philosophy first implemented in Swedish Parliament in 1997. New technologies such as driver monitoring are acknowledged in European Commission documents to hold great promise in reducing road injury. How these technologies are designed and implemented will determine whether they have minimal or maximum benefits in terms of driver support and injury reductions. The issues noted herein must be carefully considered if driver monitoring technology is to have a hope of achieving near its maximum benefit in reducing road injury.
Mike Lenné builds research partnerships with Seeing Machines’ automotive, aerospace and fleet customers. He also represents the company’s scientific interests externally and, as adjunct professor at the Monash University Accident Research Centre (MUARC) in Melbourne, maintains a high profile and considerable scientific credibility within the international human factors community. Lenné has 20 years’ experience leading applied research programs to deliver improvements in safety. Before joining Seeing Machines in 2014, he was a professor at MUARC, where for eight years he led the human factors research team. He holds a PhD in Human Factors Psychology (Monash University).