The automated driving functions with most potential: Q&A with Frederik Diederichs, Fraunhofer IAO

Add bookmark

Sumit Dutta

Autonomous cars have gone from concept to possibility in the last few years, and the companies innovating in the field are dealing with increasingly complex challenges in order to make the leap from possibility to reality.

Frederik Diederichs, Human-Factors Engineering at Fraunhofer IAO, speaks about the opportunities and possibilities ahead.

Frederik Diederichs

We find a growing number of technical system components and functionality in today’s production chassis control and driver assistance systems, which open the way to automated driving. In your view, which technologies are the biggest drivers of this?

Honestly speaking and influenced from my human factors background, the biggest driver for automated driving is actually the users’ demand for “always on” services which come along with really critical distraction issues. Car connectivity and personal devices promise on one hand productive and enjoyable travel times even in private transport but on the other hand consequently call for higher automation in order to mitigate driver distraction.

This demand hits today a technical readiness level of positioning and environment recognition that allows self- driving car demonstrators and allows us to dream about fully

automated vehicles on public roads. The reality however still requires full driver’s attention on the road and fully autonomous driving remains a vision with a horizon of 20 years.

Which automated driving functions do you think have the greatest future potential, and where do the greatest challenges lie?

Vehicle automation brings a benefit for safety, comfort and efficiency. Most automakers have highway driving automation in their roadmaps as one of the first applications. Highways offer the most structured and well maintained infrastructure with relatively low level of exceptions and a forgivable road design. Technically this scenario is one of the easiest to handle and current cars proof already the feasibility.

Currently I see the challenges here in the balance of pure supervision and active intervention and in the Human Machine Interface which needs to maintain attention, provide system transparency and support the correct mental models. Controlling and supervising automated systems is a boring and exhausting task for the human, especially over a long period.

For this reason I personally expect highly automated systems first to come in short time applications, such as emergency support, valet parking, traffic jam and traffic cueing, lane changes and merging. The challenges I see are robustness, cost and intuitive use.

How important is the development of sensor technologies, and what does Car2X offer us in terms of automated driving?

As a human factors specialist, my sensor requirements are rather directed inside the car and towards the driver state detection and the generation of driver models and profiles. Eyetracking needs to enhance robustness on the variety of users and the technology needs to become cheaper. 

Personalization based on user profiles and behavior learning also offers a high degree of individually tailored assistance. We are currently running experiments where driver states and driver models are used for negotiating with other car drivers via Car2x technology in intersections and merging situations. Also automated cars could profit from the Car2x communication in order to align or negotiate their trajectories.

What opportunities are opening up, and how do we guarantee the highest level possible of client acceptance with regard to automated driving?

Opportunities are given in all major user needs: Safety, comfort and efficiency.  Incremental  enhancements are introduced constantly. Revolutionary game changes would need to provide extra time to the chauffeurs. User acceptance depends mainly on added value and usability.

Modern user experience concepts add personal values, status message, image, joy of use, aesthetics and prizing. Automated driving functions have the potential of addressing various of those requirements with an outstanding impact and thus have a high potential for client acceptance and market success.

Considering all the prevailing conditions: Where do we stand today, and where will we stand in 2024 on the “Road to Automated Drive”?

Today we are facing a hype of automated driving which was stimulated by Google’s demonstrator cars. Most people forget that a Google self-driving car uses sensors with aprice of a Ferrari. However the OEM could also show on last year’s Frankfurt motor show and CES that automated driving is feasible with sensor technology close to mass production.

I see a strong trend towards maneuver automationwhich will in 10 years bring quite a few high class vehicles on the marked which will allow the driver toget rid of annoying tasks. In a 20 years horizon I think we will have many driving tasks automated andeven can allow drivers to read and write e-mails or whatever distracting communication mode we will want to use.

Presented in association with the Autonomous Vehicles Summit