What is hazard perception and how can online assessment and training improve fleet safety?

What is hazard perception?

Hazard perception (HP) is typically measured using a video-based test that is shown from the driver’s perspective. A driver watches a clip and presses a button when they see a ‘hazard’. This is usually defined as an event that would make the driver stop, slow down, or change direction to avoid a collision. The faster the driver responds with in a predefined window, the more points they score. Now, all of this may sound relatively simple, but HP is a multifaceted skill. Drivers must first know where to look for the most likely hazards. They must then identify ‘hazardous precursors’, such as parked vehicles that might occlude traffic emerging from a side road and place these precursors in a priority hierarchy. They must monitor all potential hazards, looking for subtle characteristics that suggest a potential hazard will develop into a real hazard. Finally, they must then respond quickly by pressing the button. All these processes contribute to how quickly a driver responds during a hazard perception test, often without explicit awareness.

What’s so special about hazard perception?

Academic research has demonstrated that HP skill is the only cognitive skill that has been consistently linked to drivers’ collision risk. These studies date back to the 60’s and there is now a large body of evidence that links HP performance to both retrospective crash risk (i.e., drivers who have previously crashed tend to perform poorer on HP tests) and prospective crash risk (i.e., drivers who perform poorly on HP tests are more likely to have a crash in the near future).

Perhaps one of the most powerful examples of this was a large-scale study conducted in Australia. In this study, they took data from 100,000 learner drivers who completed a hazard perception test and analysed their subsequent involvement in police-reported casualty collisions in their first year of driving. They found that those drivers who has scored poorly on the hazard perception test were twice as likely to have been involved in a fatal collision in their first year of driving (Drummond, 2000).

In a similar vein, numerous studies have shown novice drivers (whom we know are over-represented in crash statistics) typically perform worse on HP tests than more experienced drivers (e.g., Pelz and Krupat, 1974; Watts and Quimby, 1979; McKenna and Crick, 1991; Renge,1998; Wallis and Horswill, 2007; Horswill et al., 2008; Deery, 1999; Pradhan et al., 2009). The UK government were so convinced by this evidence that in 2002, they introduced the HP test as part of the national driver licencing procedure, meaning that all new drivers must pass this test before they can get a full driving licence. The HP test therefore keeps the worst drivers off the road while encouraging learner drivers to improve their skills. HP training has now become a thriving part of the driver training industry, though to varying levels of sophistication.

How successful has the introduction of the HP test been for improving road safety in the UK?

The introduction to the UK’s driver licencing procedure in 2002 has been a great success. A study by the Transport Research Laboratory in 2008 reported that the test lowered the collision risk of newly qualified drivers (Wells et al., 2008). More recent estimates suggest the test saves the UK nearly £90m per year by preventing over 1,000 injury collisions and more than 8,000 damage-only collisions (Horswill, 2016). This success has been acknowledged with two Prince Michael Road Safety Awards.

How is this relevant for fleet drivers?

We know that HP is good. It measures a vital skill and keep the worst drivers off the road, but how can it help improve fleet driver safety?

The official DVSA test does also form part of the Driver CPC for aspiring lorry and bus drivers. In this version, drivers see 19 clips containing 20 hazards and must score a minimum of 67/100 to pass. In comparison, learners see 14 clips (15 hazards) and have a pass mark of 44/75. The tests and clips are otherwise identical regardless of whether you are learning to drive a car, or an experienced driver seeking a professional licence for bus or HGV driving.

Beyond the need to pass the DVSA HP test to get a specialist professional licence, hazard perception training and assessment has yet to significantly impact upon fleet driver safety. There are, however, many ways in which hazard perception can benefit commercial fleets.

- - Driver recruitment: A valid HP test is a useful recruitment tool. It is often a relatively small number of drivers who account for the majority of road safety related costs in a fleet. If you can identifying these at-risk drivers before employing them, you can save a lot of money. It might seem odd to put extra barriers in the way of recruiting at a time when demand for drivers outstrips supply, but this approach will pay financial dividends in the future,
  - Identifying training needs: Some drivers need more help than others when it comes to training. Hazard perception tests can help you identify those drivers who need immediate training, and help you tailor the training to each drivers’ specific problem.
  - Online and classroom training: Mere exposure to hazards in HP tests can improve drivers’ awareness of what might happen on the road. More involved training can include explicit instruction, feedback on where (and why) drivers should be looking to spot hazards, and interactive exercises. Training can be given online via sophisticated eLearning platforms, or in classrooms with interactive activities that engage all trainees.

International research has demonstrated that hazard perception training, when done correctly, can make drivers safer. The benefits of training are not restricted to reducing collisions though. The increased awareness of hazards often results in fewer instances of harsh braking and harsh acceleration, reducing fuel consumption and vehicle wear and tear. Recent research even suggests that hazard training makes drivers less willing to engage in distracting activities, such as mobile phone use. Most distracting tasks are initiated by drivers who think the road ahead is safe. After hazard training, however, drivers become less inclined to take their attention away from the road.

But there are a few problems…

Despite the successes of the award-winning DVSA HP test (and we are huge fans!), it is not perfect. There are several outstanding issues concerning its use, particularly with professional drivers.

First, there is the question of how relevant a car-based test, which is normed on learner drivers, is to drivers of different vehicle types, e.g., HGVs, vans, motorcycles, buses or even emergency responders. If we look at HGVs for example, while every new HGV driver must pass an HP test (as part of their initial Certificate of Professional Competence training), the test is no different to that which learner car drivers must pass (albeit with a higher pass mark). This test does not reflect the hazards that HGV drivers encounter on the road. For instance, the elevated viewpoint from an HGV cab provides a better view of hazards further away from the vehicle, though hazards closer to the vehicle might be more challenging to detect. Also, the way other road users behave in the presence of an HGV is different to that of a car. We argue that HGV drivers need a hazard perception test that is specific to their vehicle and role, i.e., showing the view from an HGV on the roads and trips that an HGV is likely to travel (see Figure 1).

Another problem with the DVSA HP test became apparent when driving instructors were first required to complete the test: Many instructors score remarkably poorly! This wasn’t because they didn’t understand what a hazard was. The problem was they could predict a hazard so far in advance from very early clues that many of them would press too soon, before the scoring window had opened. This underscores the subjective nature of how the DVSA test is scored and is a problem that many aspiring HGV drivers may face, especially if they are already highly experienced car drivers.

We do not know how the UK Government decides to set the scoring windows, and many developing hazards do not lend themselves to an absolute threshold. It can be very difficult to say exactly when a potential hazard becomes an actual hazard, but the DVSA’s scoring windows are black and white in this regard: at one moment, the clue is not yet a hazard, but a split second later we should now regard this as a hazard worth responding to.

The current test also has no measure of accuracy. When a driver presses for a hazard we do not know what the driver is responding to. It is assumed that a response in the scoring window reflects detection of the developing hazard (e.g., a pedestrian stepping out from behind a parked vehicle), but it is also possible that they are pressing for some other reason than the true hazard (e.g., an oncoming car that might turn across your path).

A final problem (at least ‘final’ in terms of how many we can discuss here!) is that the nature of the hazards is constrained by the typical boundaries of a computer monitor, with hazards only ever appearing within a forward arc of approximately 60-80 degrees of visual angle. In other words, we force drivers to look in very small area of the visual world, and then make hazards pop-up in the very same area of the world. This likely inflates the hazard scores of drivers compared to what they might spot in the real world when they have 360-degree of environment to be concerned about. Such restricted hazard placement also removes the possibility of hazards appearing from the immediate left or right (e.g., as we pass by side roads), or any form of overtaking or undertaking hazard.

Tailoring hazard perception for fleet drivers?

Fortunately, advances in HP research provide solutions to the problems with the DVSA test, and have removed the barriers that may previously have deterred fleet managers from investing in hazard training and assessment. At Esitu Solutions, we believe that any hazard assessment or training should match both the vehicle and the type of driving that a driver does in real life. For that purpose, we have developed a range of different hazard tests for a range of vehicle types and job roles, including, cars, vans, HGVs, and emergency response vehicles (see Figure 1). These assessments are all filmed from these actual vehicles and therefore better reflect the type of hazards that the driver of that vehicle will encounter on a daily basis. As can be seen from Figure 1, we also include mirrors. The information in these mirrors is captured from a range of cameras attached to the film vehicle. This additional situational information makes our clips look more realistic, and they allow for overtaking or undertaking hazards to be included. In our most recent development we have gone one stage further! Using 360-degree hazard clips presented in VR headsets, we allow drivers to look wherever they want to in the world. These takes engagement and realism to the next level, which still sorting the safe from the less-safe drivers (see Figure 2).

We have even solved the problem of those pesky scoring windows. Instead of pressing a button to acknowledge a hazard, we stop the video as the hazard develops and ask, ‘What happens next?’ Trainees choose from four options displayed onscreen. If the driver has read the road appropriately, they should be able to identify the correct answer. This test is known as the “Hazard Prediction Test” and we have published research to demonstrate that this test can successfully sift safe from less-safe professional drivers (e.g., Crundall & Kroll, 2018) We have even convinced the governments in the Netherlands, Republic of Ireland and Czechia to use our hazard prediction test format, rather than that of the DVSA, in the development of their own national hazard tests.

It has taken over a decade of research, and a slew of peer-reviewed academic articles, but we believe Esitu Solutions now has an alternative hazard test that overcomes the flaws in the DVSA test and is ideally suited to fleet drivers.

What is hazard perception and how can online assessment and training improve fleet safety?