Beyond lap times: Measuring the driver’s mind with eye tracking

When it comes to analyzing the performance of a vehicle, or one of its components, in order to create an accurate virtual model, there is only one path: Measurement.

The right tools 

If we are talking about a race car, the first parameter that comes to mind is usually top speed or lap time. However, to build an accurate mathematical model many more measurements are required, and they are not always as easy to obtain as reading a speedometer or using a stopwatch. 

For instance, measuring aerodynamic load or identifying the limit beyond which tires begin to lose grip requires specific instrumentation. Vehicles must be equipped with devices such as load cells, pitot tubes, or optical speed sensors. 

In short, the right tools are needed. 

Tools capable of providing precise numbers and reliable measurements. Thanks to this data, we can, for example, compare two different tire compounds on the same vehicle, or simply evaluate the performance of two cars driven on the same circuit. 

Ultimately, mathematics, numerical analysis and, of course, the stopwatch will determine — beyond any doubt — which solution performs best. 

The human factors 

But what happens if two identical vehicles are driven by different drivers? 

How can we be certain that both drivers have reached the absolute limit? 

How can we eliminate the variable represented by the driver to ensure the reliability of our measurements? 

Or perhaps we should take the opposite approach: Measure the driver as well, turning sensations and intuition into measurable and comparable elements. 

The answer is: Maybe. 

The human factor remains, fortunately, the decisive one. Do sensors exist that can measure driving skill or, more simply, a driver’s talent or natural predisposition for speed? 

In the real world, the final judge is always the stopwatch. Under equal conditions, it is enough to determine the best among two, twenty or even two hundred drivers. Unfortunately for those working in simulation, this verdict is truly reliable only in the real world. 

In simulation environments, several factors differ from reality and make the stopwatch a less reliable judge. The perception of danger, the physical possibility of getting hurt, psychological pressure, and physical stress are all elements that influence a driver's performance in the real world but are significantly reduced in a simulator. 

The advantages of simulation  

Simulation also offers unique advantages that are rarely possible in other sports. 

For example, during a session it is possible to place biometric sensors directly on the driver and measure parameters such as: 

• Breathing rate 

• Heart rate 

• Perspiration 

• Muscular effort 

These data provide valuable information about the physical condition of the athlete. 

Unfortunately, these sensors mainly measure the driver's fitness level, which is undoubtedly important in motorsport, but they do not measure the most critical ability: The driver's capability to push a vehicle to and beyond its limits. 

Skydrive’s research 

In recent years, at Skydrive, driven by the need to evaluate and train drivers more effectively, we have gone further by researching and introducing new measurement systems. 

Among these, one of the most innovative tools we have adopted is
wearable eye trackers developed by Tobii. 

Eye tracking and driver training  

This remarkable technology provides both recorded data and real-time information, significantly enhancing our ability to analyze driver performance. At the same time, it has introduced entirely new tools for training drivers in areas that, until recently, were largely unexplored. The result has been a true qualitative leap forward in both analysis and training processes. 

An example? 

Perhaps the most striking result is that we can now observe and analyze what is commonly referred to as the “ideal racing line” inside the driver’s mind. 

Thanks to eye tracking, we can compare professional race-winning drivers with young beginners or gentleman drivers with millimetric precision, independently of lap time results, which, as mentioned earlier, are not always reliable. 

We can now analyze parameters such as: 

• Pupil size 

• Peripheral vision 

• Saccadic speed 

• Fixation time 

In other words, we can begin to measure what might be called the speed of vision. Naturally, the faster the vehicle moves, the faster the driver's visual processing must be. Being able to measure these aspects represents a major step forward in driver evaluation. 

How a driver’s vision works  

The visual process during driving is essentially an alternation between fixations and saccades. 

A fixation occurs when the driver’s gaze reaches a reference point and remains focused there long enough to perform a driving task. For example, a braking reference point. 

A saccade occurs when the eye leaves that fixation and rapidly moves towards the next reference point. Let us consider the example of a braking zone followed by corner entry. Ideally, the driver identifies the braking marker and keeps the gaze fixed on it until the moment braking must begin. 

Once braking has started, the driver must determine where to direct the car for the correct corner entry. At this point, the eye performs a saccade, searching for the next reference point, often the apex of the corner, effectively drawing the ideal trajectory inside the driver’s mind. Once the apex is identified, the gaze stabilizes again in a fixation, and the driver performs the steering and vehicle control inputs necessary to guide the car precisely towards that point. 

Training the eyes  

The most interesting aspect is that this entire process can be trained. 

Drivers can train: 

• Eye movements 

• Fixation accuracy 

• Saccadic speed 

• The overall speed of visual processing 

An extraordinary performance, but one that can be approached through proper training. Considering that the eye is the fastest muscle in the human body, capable of contracting in less than 1/100 of a second, measuring these eye movements with precision is not a trivial task. 

The analytical software provided by Tobii allows us to reach exactly this level of measurement. And from that point on, the real exploration can begin. Because once something can be measured, it can also be trained. And once it can be trained, it can be improved. 

We were then able to begin identifying and developing specific metrics, which are fundamental for parameterizing driver performance. Once these metrics had been defined, we collected reference samples by working with professional drivers who agreed to undergo a series of tests at the wheel of our simulator and driving the real car as well. 

Turning vision into performance 

After developing a system capable of reproducing and displaying the same metrics in a clear and intuitive way, we invited less experienced drivers — such as young talents or amateur drivers — to undergo the same tests and driving exercises. 

The next step was a natural one:

Comparing the two sets of metrics. Initially, this comparison was carried out in a static way, after the driving session on the simulator. This allowed the less experienced driver to clearly understand the gap between their performance and that of a professional driver, enabling them to make adjustments and improvements in the following session. 

Later, however, the system was implemented with real-time graphical visualization on screen. The driver undergoing training can now see projected in front of them the points that represent the professional driver's fixations and saccades, making it easier to follow those movements and train their own visual process. 

At first this process may feel somewhat mechanical. However, with repeated training sessions, the visual behavior of the AM driver progressively improves, gradually becoming a subconscious and natural process, ultimately allowing the driver to acquire a genuinely higher level of driving skill. 

This powerful combination of cutting-edge technology from Skydrive and Tobii unlocks the hidden potential in every next-generation driver.