Zum Hauptinhalt springen
TU Graz/ TU Graz/ Services/ News+Stories/

Do We Trust the Invisible Assistant?

12/28/2016 | Planet research | FoE Mobility & Production

By Birgit Baustädter

In the MueGen Driving project, researchers from the Institute of Automotive Engineering at TU Graz studied how gender and age influence our trust in driver assistance systems.

A new driving simulator was developed for the MueGen Driving project.
“Although this is a bit of a sweeping statement, we know that young men are more likely to be involved in accidents than women of advanced age,” explains Arno Eichberger. It may sound here like a sweeping statement, but finding it to be corroborated by actual accident statistics Eichberger and colleague Cornelia Lex from TU Graz’s Institute of Automotive Engineering decided to start the MueGen Driving project. In their work, both Arno Eichberger and Cornelia Lex intensively investigate driver assistance systems that either relieve the person behind the steering wheel of a vehicle of bothersome/monotonous tasks or are ready to intervene in emergency situations and prevent dramatically worse outcomes. “If the cause of the accident really is influenced by the driver’s age and gender, the assistance systems in the vehicles must also be capable of reacting differently. The purpose of this differentiated approach must be to ensure that the driver feels at ease and accepts the system’s driving behaviour,” says Eichberger about his original research idea.

Man-machine interaction

Even though a driver assistance system supports the driver on the road, the ultimate control of the vehicle still lies with the human driver. Cornelia Lex adds: “We are dealing with the entire range of human behaviour here. This is why it is important that driver assistance systems do not stress out the human driver and take into account personal modes of behaviour.” The human driver’s trust is crucial for an optimal interaction between the human driver and the driver assistance systems, and for an optimal task execution by the system. For example, with the “adaptive cruise control system (ACC)” the car automatically selects the appropriate speed to keep a safe distance to the vehicle in front. But the driver’s judgement whether this distance is big enough, and whether the vehicle puts on the brakes at the right time, is a subjective perception which in turn strongly affects the trust in the system. 
In the MueGen Driving project a Automated Emergency Braking System and a adaptive cruise control system were tested.

MueGen Driving

Starting with this premise, the MueGen Driving project, subsidised through the FEMtech tendering channel of the Austrian Research Promotion Agency (FFG), investigated how drivers use the assistance systems differently and whether, up to which point and when they trust them. The basic parameters contemplated by the study were age, gender and driving activity. “Of course we could have looked at more parameters, but then we would have needed a very high number of test persons to come up with meaningful results,” comments Eichberger. From the very beginning, the project was competently supported by human factors expert Ioana Koglbauer from the Institute of Mechanics, a psychologist specialised in the interaction between man and machine. In her line of research it has become established practice to carry out usability tests with representative user groups: “It is imperative to gain knowledge about the perceived advantages and disadvantages of the application of technologies by all relevant customer groups.”

It is imperative to gain knowledge about the perceived advantages and disadvantages of the application of technologies by all relevant customer groups.

The study focused on two assistance systems: the safety-relevant Automated Emergency Braking System and the comfort-enhancing adaptive cruise control system. “It was our specific intention to start with simple systems that are already available as standard features in vehicles,” says Eichberger. The investigative approach started with an observation of the test persons in a real environment. The participants were asked to drive a series vehicle equipped with these systems – an adapted Audi A6 – on the road. “Of course you have to do this systematically. There is no comparative value or meaning in having people drive around just like that,” says Eichberger. “Therefore, we looked for other tests with the assistance systems and selected typical driving manoeuvres from the specifications, told the test persons to carry these out and then proceeded with the evaluation.” The input for this evaluation came firstly from a questionnaire presented to the test persons, and secondly from objective measured data such as speed, acceleration and similar. “For example we asked the test persons to objectively assess the distance to the car in front, then we compared their statement with the objective data,” Eichberger adds. 
Research was done in a real environment as well as in a driving simulator. 

All senses deceived

The second part of the research project called for a new development. If the test involves a much larger group of test persons, a real-life test is a time-consuming, risky and costly exercise. Therefore, the researchers decided to develop their own driving simulator at TU Graz – a task that took them one and a half years – to be able to carry out the experiment with a larger set of test persons under always identical conditions. “It is not that such a simulator is not available off the shelf but we thought this would have been too boring, and of course if you choose this shortcut you cannot install extensions specifically for additional research projects later,” Eichberger grins. For the simulator they converted a Mini Countryman, integrated it in a completely darkened and acoustically insulated box, and fitted monitors showing the simulated road on all windows. 
The driving simulator which was developed at TU Graz.
As a world first, they developed the autostereoscopic visualisation – a 3D visualisation that works without requiring specific 3D glasses – in cooperation with Fraunhofer AG. However, they could not use the visualisation much so far. “Unfortunately, not all boundary conditions were optimal during this first test, and so we produced a lot of simulator sickness. Nevertheless, I definitely want to continue working on this in future. However, for the study we ultimately resorted to 2D simulations.” The perfect sound and driving dynamics that make the simulation that much more realistic were provided by AVL. “We have to deceive every human sense to make the results translatable to reality,” says Eichberger to justify this immense task. This is why even the steering wheel provides feedback – as if you were driving on a real road. Cooperation partner Steer-by-Wire Technology was in charge of the time-consuming development of the feedback mechanism, which they designed for future use in a vehicle prototype. “It was very important for us to find out how the person holding the steering wheel responds to a change in the road conditions.”  
A research vehicle from interior view.
After announcements in the media and targeted public relations work, 96 test persons were selected for participation in the study. “Our youngest test person was 18 years old, our oldest at around 86.” Cornelia Lex tells us. These test persons executed the very same manoeuvres on the simulator that the test persons had previously performed in real traffic. The results were enlightening. “There really are statistically significant gender and agespecific differences that can now serve as input for the further development of the assistance systems,” Eichberger announces. The concrete results indicate that the subjective trust in automatic cruise control is comparable in men and women in the 20-29 age  bracket. With increasing age of the test persons, in women the trust level goes down while in men it goes up. It is only in the 50-59 age bracket that women start to have a high level of trust in this assistance system, but men less so. In the over-60 age bracket, the differences in trust between men and women tend to disappear, but in any case are higher than in young participants. The driving speed with the adaptive cruise control system is significantly influenced by the road condition and the driver’s age. Test persons aged 60 and above prefer a significantly lower speed than younger persons. Women of all age groups subjectively trust the Automated Emergency Braking System less than men. Also, women rate the comfort of the assistant lower than men.
“But in my opinion the biggest difference is to be found in how the test persons rate the ‘workplace’ – how information is presented and which operating options are offered,” comments Eichberger. Men respond more favourably than women to the abundance of display features that are characteristic of design of the tested series vehicles.

A glimpse of the future

“It would be possible to make assistance systems react differently, depending on whether the person behind the wheel is a man, a woman, an older or a younger person,” Eichberger says looking ahead to future applications of the research results. He adds: “A higher level of trust will also push up the acceptance and satisfaction ratings, which will automatically benefit safety.” All participants in the research project are already working on follow-up projects that study the subject in even greater depth.
The work group Driver Assistance, Vehicle Dynamics and Suspension at the Institute of Automotive Engineering developed the simulator single-handedly.

This research area is anchored in the Field of Expertise "Mobility & Production", one of five research foci of TU Graz.

Contact

Arno EICHBERGER
Assoc.Prof. Dipl.-Ing. Dr.techn.
Institute of Automotive Engineering
Inffeldgasse 11/II
8010 Graz
Phone.: +43 316 873 35210
arno.eichberger@tugraz.at