Australia – Last November, at the Euro Bus Expo in Birmingham, UK, I had the dubious pleasure of trying out a product by Seeing Machines. I was sat down in a bus driver’s seat facing a monitor and a camera. Told to let my mind wander, my eyes started to lose focus… In a split second, I was jerked rudely awake as the seat vibrated beneath me. 

I had deliberately allowed myself to become inattentive; the driver of a bus or truck keeps his attention fixed on the road ahead and on road users and obstacles to either side of him. But the representatives from Seeing Machines told me that a tired driver can fall into repeated episodes of so-called “microsleep”, a moment of sleep or drowsiness lasting no more than a few seconds which can have devastating consequences in a job that requires constant alertness. Recent history furnishes us with plenty of tragic examples of collisions caused by sleepy drivers on motorways. The National Highway Traffic Safety Administration estimates that in 2017, 91,000 police-reported crashes involved drowsy drivers and led to approximately 50,000 injuries and nearly 800 fatalities in the United States. 

Seeing Machines has been working on this problem for over two decades, and the fruit of its efforts is ‘Guardian’, a driver monitoring system which has demonstrated a more than 90% reduction in fatigue-related driving risks through product testing. This AI-driven software uses facial cues to determine whether the driver has become inattentive and employs two kinds of alert – an audio alarm and seat vibrations – to bring the driver’s focus back to the road. During the product demonstration at the Euro Bus Expo, the camera trained on my face was picking up signs of distraction – a drooping face, wandering eyes – and feeding these to the software for split-second analysis and evaluation. In principle, the technology is identical to the vision-perception systems employed by self-driving vehicles, except that the camera is turned inwards, and the software is designed to pick up on slight changes in facial expression. 

Intrigued by the product, I was determined to find out more about the company. The following is based on answers to queries sent over email.

Background to Seeing Machines

As is usual for tech firms, Seeing Machines had its start on a university campus. More unusual in the tech world is the fact that the company began not in Silicon Valley or another metropolitan “tech hub” but in the modestly sized capital city of Australia, Canberra. It was spun out from the Australian National University in 2000, launched originally in conjunction with Volvo Technological Development. The company’s vision was to enable the development of the “crash-proof vehicle” through driver monitoring technology, and it has stuck to this goal over 23 years. 

The remoteness of Australia from the developed countries of the northern hemisphere posed a challenge from the outset. “There is no automotive manufacturing industry in Australia any longer,” Seeing Machines admitted, which necessitated the company to reach out across time zones to original equipment manufacturers located far away. As such, the company has steadily grown its presence abroad in Europe, the USA, and Japan. Even so, Seeing Machines’ location has had its advantages. The company has found that it is “easy to attract” international talent to come and work in Australia and benefit from its high quality of life. The country’s stringent travel restrictions during the COVID-19 pandemic stalled its hiring efforts somewhat, but Seeing Machines says it is now “back on track”. 

Australia is a continent with a vast wilderness interior, and its urban centres on the coast are separated by many miles of road. The country’s reliance on goods transportation by truck – often in the form of “road trains” or three or four semi-trailers attached to a single prime mover – is greater than most other developed continents, where there is a mixture of railways and waterways in addition to roads on which goods can be moved. The truck is therefore a vehicle of outsize importance to the Australian economy. Moreover, Australia has one of the largest national mining sectors in the world, employing thousands of heavy-duty trucks and machinery in mineral extraction. Naturally, it was mining which provided Seeing Machines with its first commercial breakthrough in 2015. BHP Group, one of the major multinational mining corporations based in Australia, sought Seeing Machines’ help with reducing the risk of accidents involving drivers on the job. 

Driving a mining truck is an incredibly repetitive task. All through the day, the same driver must cover the same route, loading and unloading, in a hazardous environment. Even with regular breaks, the risk of accident from fatigue or distraction in such a job is significant. Seeing Machines developed and tested its ‘Driver Safety System’ through applications on mining sites, and its work paid off through a major deal with heavy equipment manufacturer Caterpillar Inc. In 2016, the American company acquired the exclusive worldwide licence for the Driver Safety System in mining applications for USD16m, and to this day, Caterpillar sells ruggedised versions of the product through its global distribution network.

Guardian in-vehicle hardware

The technology behind ‘Guardian’

Seeing Machines’ product offering for commercial fleets, called ‘Guardian’, is based on the Driver Safety System it developed for mining applications. The system comprises a computer with embedded software, a camera, and alert functions. Tracking the driver’s eye movements and head position, the artificial intelligence built into the system determines whether the driver is drowsy or distracted. When certain safety parameters are exceeded, the driver is alerted through sound alarms and seat vibrations. Each fatigue or distraction “event” is logged by the system and relayed to Seeing Machines’ technical support centre. Camera footage from every recorded event can be reviewed by fleet managers through an online platform called ‘Guardian Live’. Managers are able to monitor driver behaviours and take action if deemed necessary. Thus, Guardian not only intervenes to mitigate the risks from distracted driving, but it also allows for surveillance and correction of patterns of driving behaviour that lead to recurrent inattentiveness. 

Making sure that Guardian is able to correctly identify and respond to fatigue or distraction events is an “extremely complex” task, according to Seeing Machines. External signs of tiredness or inattentiveness may appear vastly different from one driver to another. Moreover, a poorly designed system might respond repeatedly to false signs of fatigue or distraction, thus eroding the driver’s trust in its efficacy. Ensuring that Guardian works correctly every time has required the collection of huge amounts of data to train the artificial intelligence thoroughly. As of writing this, Seeing Machines claims its driver monitoring systems have collectively travelled over 13 billion kilometres, amassing tons of real-world driving data. 

Furthermore, Seeing Machines employs a team of specialists in “human factors engineering”, an applied field of psychology which seeks to reduce human error and increase safety in interactions between humans and machines. In an informational video, Professor Mike Lenné said: “Seeing Machines is unique in that it recognised […] that human behaviour is complex and challenging, and that we needed human factors research to understand differences between driving a car, a truck, or flying a plane.” As such, the company has invested steadily in research into this interdisciplinary field in order to better understand the myriad ways in which human drivers exhibit inattentiveness. 

The software underpinning Guardian is embedded onto a range of different platforms, but the primary one is the ‘Fovio’ chip which is made by American semiconductor company Xilinx. With Seeing Machines’ ‘Occula® Neural Processing Unit’ embedded in the system, the Fovio chip’s computational load is reduced by 50% without compromising its functionality or performance. Occula is available for licence in ASIC form for integration in any automotive computer platform, as is Seeing Machines’ software library (its ‘eDME’ or ‘embedded Driver Monitoring Engine’), allowing other companies to employ and build upon its technology. 

Driver monitoring systems and the CV industry

Seeing Machines markets its products to automotive manufacturers, the aviation sector, and semiconductor companies. But the changing legal context within the European Union has led to a greater focus on the commercial vehicle industry. 

As of 2022, the EU’s General Safety Regulation already requires that new vehicle types, whether cars or CVs, be equipped with driver drowsiness and attention warning systems. From July 2024, a new set of rules is set to come into effect. Newly registered vehicles will require devices for intelligent speed assistance, reversing detection, event data recording, and emergency stop signalling. In addition, commercial vehicles will require blind-spot detection, collision warning, and tyre pressure monitoring systems. Bundled together with this raft of new requirements is driver monitoring systems. In 2026, new vehicles will require more sophisticated systems capable of tracking driver (in)attention. 

Seeing Machines already has partnerships with automotive OEMs such as Ford, Mercedes-Benz, and General Motors. At the Euro Bus Expo, I was told of discussions with Switch Mobility and Tevva. 

Seeing Machines was founded with the aim of reducing collisions caused by driver inattentiveness, and this remains the company’s focus. Naturally, it welcomes legislative efforts in Europe to enforce the adoption of advanced systems for mitigating the risks caused by driver fatigue and distraction. Seeing Machines says it is working hard to convince the authorities in Australia to follow Europe’s lead: “we think Australia needs to get a move on!” 

The company provided this justification for its technology: 

Humans are responsible for most road related accidents. We believe that assisting humans to be more attentive and aware of road conditions, their own conditions and careful when using ‘assisted driving’ technologies, is the ultimate and only way to improved safety on roads. But this means providing the relevant technologies to all road users, in all countries.

To this end, Seeing Machines has worked over several decades to make a product which is fit for the job. As new regulations requiring the use of advanced driver monitoring systems come into effect, vehicle manufacturers will need to pay more attention.