Eye tracking for driver safety

Audi's Attitudes, a corporate social responsibility program, wanted to learn more about the phenomenon of driving without awareness. Wearable eye trackers from Tobii Pro enabled their researchers to gain insight into driving behavior and occurrences that are often forgotten by drivers. The data revealed significant changes in gaze patterns during phases of subconscious driving and helped identify situations and external factors causing the issue.

Background to driving study

DWA(Driving without awareness - 무의식 운전)는 주변 교통에 대한 의식적인 주의가 거의 없는 상태에서 차량을 운전하는 것으로, 잠재의식 운전이라고도 합니다. 이 당시에는 이러한 주제에 대한 연구가 거의 없었습니다. 과거의 연구는 모두 긴 고속도로 여행에 초점을 맞추었고, 실제 주행 경험에서 벗어나는 시뮬레이터를 사용했으며, 운전자를 인터뷰하는 방법을 사용했습니다.
 
이 연구의 목적은 DWA에 대해 자세히 연구함으로써 이러한 방식으로 운전하는 운전자의 특성을 정확히 파악하고, DWA의 발생을 촉진할 수 있는 상황을 파악하는 데 있었습니다. 또한 이 문제에 대한 운전자의 인식을 조사하고 도로 안전에 미치는 영향을 분석하는 것을 목표로 했습니다.

시선추적을 통해 실시간으로 상세하게 운전자의 주행 행동을 연구할 수 있었고, 무의식적인 주행 상황을 파악할 수 있었습니다.

Hildebrand Salvat,, Partner and Technical Director at Gabinet Ceres, SL

Tools and methods

A man driving an Audi car wearing Tobii Pro Glasses eye tracker.

연구는 두 단계로 수행되었습니다. 첫 번째, 2000명의 스페인 운전자를  대상으로 전화 설문 조사를 실시하여 DWA의 다양한 측면에 대해 물었습니다. 두 번째 단계에서는 시선추적 연구를 심층 인터뷰와 결합하여 몇 가지 정의된 가설을 테스트하고 무의식적인 운전에 대해 더 자세히 조사했습니다.

테스트된 가설 중 몇 가지는 다음과 같습니다.

  • 만약 운전자가 어떤 생각이나 생각에 집중하고 있다면, 무의식적인 운전은 더 일어날 가능성이 높습니다.
  • 익숙한 길을 지날 땐 자극에 덜 주의를 기울이게 됩니다. 
  • 많은 자극이 있으면 무의식적 운전을 할 가능성이 줄어듭니다.
  • 운전자는  운전 행위와 무관한 자극보다 관련된 자극에 더 관심이 높습니다. 

시선추적을 통한 관찰

13명의 숙련된 운전사들, 같은 산업지역과 25세에서 60세 사이의 모든 노동자들이 각각 별도의 목표를 가진 두 가지 다른 시험 동안 관찰되었습니다. 첫 번째 테스트에서 참가자들은 경주 트랙 회로를 운전했습니다. 그래서 연구원들은 유사한 교통 조건에서 참가자들을 비교하면서 각 랩의 고유한 특성을 사용하여 가설을 테스트할 수 있었습니다.

두 번째 테스트에서, 운전자들은 대부분 출퇴근을 위해 정기적으로 다니던 시외 도로를 운전했습니다. 이 테스트를 통해 연구자들은 실제 상황에서 무의식적인 운전을 분석하고 실제 주행 상황에 대해 보다 일반화할 수 있는 결론을 도출할 수 있었습니다.

A test participant wearing Tobii Pro Glasses eye tracker.
A man driving an Audi car wearing Tobii Pro Glasses.

During both tests, participants wore wearable eye trackers from Tobii Pro to record their eye movements, enabling researchers to determine which items in a participant's field of vision attracted their attention. After each trip, in-depth interviews were conducted with the drivers. The participants did not know or were not told the purpose of the study. They were asked what they remembered seeing along the road, what they thought they had done behind the wheel, and what they normally do while driving.

Analysis of eye tracking and interview data

By analyzing eye tracking data and mapping it against verbal feedback from the participants, researchers were able to gain insight into the DWA phenomenon. To test the hypotheses, they examined relevant gaze replays from all of the trips. Tobii Pro Studio was used to create visualizations such as heat maps and gaze plots to illustrate the drivers' looking patterns and what they had actually seen.

Results

The Attitudes study found that DWA is widespread as a result of experience and driving practice. Conversely, in the survey, only 25% answered that they had never experienced subconscious driving and 5% reported experiencing such situations on a regular basis. Furthermore, the study showed that motorists who experience DWA are typically seasoned drivers who use the car daily to get to work or school, usually traveling the same route.

In addition, the research showed evidence of DWA among the drivers who said they drove subconsciously every day and among drivers who said they mostly were aware of everything as they drove. It is therefore reasonable to assume that not all drivers realize this issue is happening.

Is driving without awareness safe driving?

The study concludes that DWA reduces attention to the act of driving, but not necessarily to general security and road safety. Should something occur that is perceived as dangerous, subconscious driving disappears and most drivers can react appropriately. Researchers were unable to test effects on reaction time, but estimated that an increase would most likely be very small. Driving without awareness can be considered a moderate risk, but may have a potentiating effect on accidents when combined with other factors such as excessive speeds.

Significant changes in gaze patterns

In conscious driving, the driver's gaze remains fairly constant and identifies relevant stimuli without any substantial jumping from one point to another. Eye tracking data revealed significant changes in gaze patterns during phases of subconscious driving. An increased number of fixations, an expansion of the viewing area, and more rapid eye movements were observed although the driver's gaze remained focused on a particular area of the road. When driving subconsciously, participants used what the researchers called 'active radar vision,' which covers a broad field while drivers remain alert to the possible occurrence of stimuli.

Gaze plots illustrating attention pattern while driving a car.

Eye tracking data also showed that attention to relevant orientation points declines when driving subconsciously, but drivers still look for potential risks. They look less at stimuli that do not vary, which are predictable and known. The drivers perform appropriate maneuvers but do not remember most of them when asked. Control actions, such as checking the speedometer when passing a new speed limit sign, decreased during DWA.

Parallel tasks and thinking

The study indicated that DWA is more likely to occur when the driving is done in combination with other tasks. All participants who experienced phases of subconscious driving explained later that they had been thinking about other things not related to their driving, such as organizing their work day or decisions they needed to make.

Gaze plots and heat maps illustrating attention pattern while driving a car.

Familiarization with the route

DWA is more likely to occur on familiar routes, where participants said they can drive in a more relaxed way and pay less attention to stimuli along the road.

Eye tracking data shows that when traveling on familiar roads, most drivers always attend to stimuli that vary, while a significant number pay less attention to stimuli that are constants. Familiarization with the route caused a decrease in both the driver's viewing area and the number of fixations and eye movements. The gaze plot and heat map visualizations below show differences in gaze patterns during the first and last lap on the race track circuit when participants knew the layout better.

Number of stimuli on the road

Boring routes with less stimuli are another factor that can promote DWA. Eye tracking data shows a correlation between the moment when a driver fails to notice certain stimuli on the road and sections in which the number of stimuli is lower. Absence of stimuli does not necessarily mean DWA will occur, but when the route is unknown and has many stimuli, subconscious driving is less likely.

Relevant and irrelevant stimuli for driving

Results indicate that, when driving subconsciously, drivers can see most of the relevant stimuli that appear in their path, even with a wider range of coverage, and that they are prepared to detect relevant stimuli. Eye tracking data verifies that they pay more attention to relevant stimuli (speed limit signs, traffic lights, etc.) than to irrelevant stimuli (ads and banners). Drivers do not always remember what they saw or what actions they took as a result. While many motorists do not see irrelevant stimuli, they often remembered by the drivers who have seen them.

Gaze plots illustrating attention pattern while driving a car.

Cognitive resources savings

The fact that drivers do not remember everything that they saw or that occurred during the journey reflects the role of cognitive resources savings. Although they pay attention at the time, they do not allocate resources to store this information for later recall because it is not considered necessary. However, they may remember some stimuli that may be conspicuous but irrelevant or even outside traffic, or they recall maneuvers made by other drivers. This reinforces the idea that subconscious driving is restricted only to control repetitive actions known to the driver himself.

Related information