Improving Driver Safety Training and Assessment

In 2018-2019, 27,820 people were killed or seriously injured (KSI) on British roads. NTU researchers found that identifying hazardous precursors is a vital safety-related skill that likely underpins many such KSIs. This led to the development and implementation of award-winning hazard tests for fire-service drivers, a hazard test for a national bus operator, VR hazard training for 25,000 London bus drivers, and a free-to-download VR app for learner drivers. NTU advised the Dutch and Irish Governments on the development of national hazard tests and has licensed tests to a leading health & safety e-learning provider (Human Focus International).

NTU researchers have concluded that predicting imminent hazards based on hazardous precursors (i.e. clues in the driving scene) is the vital skill that makes drivers safe, rather than simply reacting quickly to hazards. This finding underpinned development of more robust tests of driver safety that are preferred by drivers, viewed as fairer tests and are better suited to professional and international driving contexts.

NTU’s new tests are based on hazard prediction: Drivers watch clips of driving which occlude just as the hazard begins. When asked “What happens next?”, they choose an answer from on-screen options. This test mitigates flaws in the official UK hazard perception test, including scoring subjectivity and criterion bias. Based on the Situation Awareness Global Assessment Technique, this new test has fared better than more traditional hazard perception tests in differentiating safe and less-safe drivers in many circumstances.

These insights arose from research, including externally-funded projects, that created and validated tests to differentiate safe from less-safe drivers, refining protocols for test development and expanding understanding of the underlying skill.

Two grants from the Fire Service Research and Training Trust funded development of video-based tests for fire drivers, with clips filmed from fire engines and cars on blue-light runs. This was the first study to directly compare prediction and perception test-formats, with the former found to be more successful. The resultant tests are now used in fire service driver training and are freely available on testmydriving.com. This research won the Fire Magazine research award and led to three publications.

The Department for Transport funded research to identify which scenarios best suit the prediction or perception test-format using a novel hazard test constructed from Computer Generated Imagery. This test had several innovations, such as driving theory questions embedded in the clips. The tests were well received, successfully differentiated between safe and less-safe driver groups, resulting in a publication.

Several other grants have allowed NTU to validate and iterate the test-development process for specialist populations. Funders include Stagecoach, to develop a hazard test for bus drivers, the Road Safety Trust (RST), to develop a hazard training for mobility scooter users, and a French company (IMRA), to develop hazard tests for autonomous vehicle algorithms. The findings from these projects have refined NTU protocols for hazard assessment and training.

PhD research, supported by small grants from Santander and the AngloIsraeli Association, has confirmed the appropriateness of the hazard prediction approach for international audiences while also further refining the methodology.

The most recent grant has allowed NTU to develop 360-degree hazard tests for virtual reality (VR) headsets. In comparisons with traditional tests, VR is more effective for assessment and training purposes. The project also led to the launch of ‘Hazard Perception VR’ in the Oculus Store, and the subsequent incorporation of a University spinout company (Esitu Solutions Ltd.) with funding from Innovate UK and investment from NTU.

Road traffic collisions are the 8th leading cause of global death and are predicted to rise to the 5th leading cause by 2030. Though Great Britain is considered one of the safest countries for driving, there were still 1,752 fatalities in 2018-19 and nearly 26,000 serious injuries. Two driver groups are of particular concern: young drivers who are over-represented in collisions (3x more likely to have a crash than older drivers), and commercial drivers (involved in approximately 30% of fatal collisions). These data contextualise the need for better training and assessment of hazard avoidance skills, acknowledged to be a leading cause of such collisions.

NTU’s impact has been to change the paradigms and resources used to train and assess the hazard avoidance skills of various categories of road user, developing and promulgating evidencebased techniques with the aim of a longer-term impact on collision statistics.

In regard to impact on learner driver hazard assessment and training, TSO (formally The Stationary Office) asked Professor David Crundall to contribute to the Learner Driver DVD, which is sold nationwide as a resource for drivers preparing to take their test. Based on their work in hazard perception, Crundall and Professor Groeger were then funded by the Department for Transport to develop a novel and innovative hazard perception test. The report recommended:

1. longer and more complex hazard clips
2. combining theory knowledge and hazard awareness in a single test
3. introducing hazard prediction clips into a future test

Based on these results, Crundall was called to advise the DVSA, and their counterparts in Ireland (The Road Safety Authority) and the Netherlands (the Centraal Bureau Rijvaardigheidsbewijzen). The RSA is pushing forwards with a hazard test, based primarily on Crundall’s data. The CBR were already planning a national hazard perception test, but they have changed this to a hazard prediction test based on evidence presented by Crundall. Meanwhile, in the UK, the importance of our research to UK national test development is acknowledged in the Government’s 2019 Road Safety Statement. Most recently, externally-funded research has allowed the release of an Oculus app designed for learner drivers entitled ‘Hazard Perception VR’. Launched in November 2020 it has been well received by users and has received media coverage from road safety outlets.

The second target impact group was professional drivers. Crundall and Brunsden were funded to develop a hazard perception test for fire-appliance drivers by the Fire Service Research and Training Trust, leading to three publications. The award-winning resources (FIRE Magazine 2017) are now being used by Nottingham Fire and Rescue Service (NFRS) to train new and existing fire drivers and have been made available nationally via testmydriving.com. A recent NFRS update includes positive feedback from their trainers including, “The “What happens next” section is good for drivers at all stages in their development and encourages them to re-examine their perceptions of everyday driving hazards”.

Finally, Crundall was asked to help develop a VR hazard prediction training tool for 25,000 London bus drivers on behalf of Transport for London. Working with STEPS Drama (a London-based training company), Crundall designed the VR assessment and training package, building on NTU published outputs and new insights from recently funded VR research. Entitled Destination Zero, the training package was commissioned as part of the Mayor of London’s drive towards ‘Vision zero’ with the aim of zero bus-related fatalities by 2030.

The success of the impact to date resulted in the creation of a spinout company in July 2020, funded by Innovate UK and a further investment from NTU. With one full-time employee and two part-time employees, the company aims to commercialise the research assets and knowhow developed by the group.

• Crundall, D. (2016). Hazard prediction discriminates between novice and experienced drivers. Accident Analysis and Prevention, 86, 47-58. doi: 10.1016/j.aap.2015.10.006
• Crundall, D., and Kroll, V. (2018). Prediction and perception of hazards in professional drivers: Does hazard perception skill differ between safe and less-safe fire appliance drivers? Accident Analysis and Prevention, 121, 335-346. doi: 10.1016/j.aap.2018.05.013
• Ventsislavova, P., Crundall, D., Baguley, T., Castro, C., Gugliotta, A., GarciaFernandez, P., Zhang, W., Ba, Y., and Li, Q. (2019). A comparison of hazard perception and hazard prediction tests across China, Spain and the UK. Accident Analysis and Prevention, 122, 268-286. doi 10.1016/j.aap.2018.10.010
• Ventsislavova, P., and Crundall, D. (2018). The hazard prediction test: a comparison of free-response and multiple-choice formats. Safety Science, 109, 246- 255. doi: 10.1016/j.ssci.2018.06.004
• Crundall, D., van Loon, E., Baguley, T., and Kroll, V., (2021; available online 18 Nov, 2020). A novel driving assessment combining hazard perception, hazard prediction and theory questions. Accident Analysis and Prevention, 149, 105847. doi. 10.1016/j.aap.2020.105847
• Kroll, V. R., Mackenzie, A. K., Goodge, T., Hill, R., Davies, R. L., and Crundall, D. (2020). Creating a hazard-based training and assessment tool for emergency response drivers Accident Analysis and Prevention, 144, 105607. doi.org/10.1016/j.aap.2020.105607