An interview with Dr. Oliver Bühler and Dr. Daniel Ulmer, managing directors of Steinbeis Interagierende Systeme GmbH
Will we still be driving cars in the future or will cars drive us? To answer this question, a colossal volume of test data has to be gathered and evaluated. The Steinbeis experts Dr. Oliver Bühler and Dr. Daniel Ulmer explain how this could be done and talk about the influence car data will have on our driving habits in the future.
Dr. Bühler, Dr. Ulmer, you’ve been testing driver assistance systems and developing embedded systems since 2005. Building on this expertise, in 2012 you founded Steinbeis Interagierende Systeme GmbH. Your work involves understanding huge volumes of data and extremely complex algorithms – safely and with certainty, efficiently, and effectively. What is it that fascinates you so much?
Oliver Bühler: With simulations, you can look at driver assistance systems in a completely different way compared to real vehicles. You do need to keep in mind that a modeled environment will trigger a somewhat different reaction from the assistance system during a simulation than in a real vehicle, but there are pretty efficient ways to examine this behavior in different situations and this can be automatically pushed to the limit. What impresses me is the way after 36 hours of simulation, harmlessly steering away from something can lead to a critical situation because the computer has been used to probe the thresholds of the driver assistance system. We’re dealing with the very latest technologies here, not just from the point of view of testing but also in terms of our own software developments, because we’re trying to come up with the best possible test results with the resources that are available to us.
Advanced driver assistance systems can now not just make an essential contribution to driving comfort, safety, and energy efficiency, they can also proactively help with driving. What will the consequences of such developments be for the travel and transportation industry?
Daniel Ulmer: The driver should become redundant. Studies have shown that young people would much rather play around with their smartphones than with cars, which have to be repaired, filled up with gas, and need 100% attention to drive. The moment the driver assistance system is intelligent enough to take over part of the drive, drivers can spend more time with their smartphones and the vehicle will even look after filling up with fuel and finding a parking space. Autonomous vehicles retain the appeal of cars for people who are actually only interested in getting from A to B and aren’t necessarily that interested in the car itself.
Before people actually start using driver assistance systems and cars start driving autonomously, a tremendous amount of simulation is needed, involving gargantuan volumes of data. What implications does this have for the kinds of “driver substitution systems” people are already talking about, especially if we want to achieve the same levels of safety as autonomous vehicles with driver assistance systems?
Oliver Bühler: One important factor is lightening the burden on the driver and how we actually convey this idea of lightening the load. When they first introduced cruise control buttons, or speed regulators, the idea was to just relieve the driver of the task of pressing the gas pedal, but even then people misinterpreted it and there were accidents because some drivers thought they could leave the vehicle to drive without supervision. So this means the difference will be how much responsibility lies with the carmakers in allowing their customers to take their eyes off the road. If a car company wants to be sure that the vehicle is reacting properly in different kinds of situations and environments, and that the driver is allowed to take his or her eyes off the road, given the current state of technology this will only be possible by doing additional simulations and going through as many realistic scenarios as possible. That might sound easy, but the main factor driving the exploding volume of simulations is where and when the carmakers are going to be responsible, so in which environments and for how long their software will drive the vehicle.
Dr. Ulmer, what’s your take on this question: In the future, will the data held by a car have to put up with drivers, will the drivers feel put out by the car driving itself, or will it be more the case that drivers actually use data for their own travel requirements?
I’m sure it will be the latter case. I say this because when you network vehicles, the data that’s transmitted can be used to improve route planning and warn drivers of dangers. On the one hand this is because if data can’t be used, the way things are at the moment autonomous driving wouldn’t be possible, and on the other hand customers derive immediate value from the data.
Gathering data is an opportunity but it’s also a threat, because any data that exists can be analyzed. Taking that as the basis, we all have to decide for ourselves to what extent we want to influence or dictate the way the vehicle is driven. It’s quite conceivable that one day it will be possible to generate much more detailed and much more objective reconstructions of what’s happening on the roads.
In technical terms, autonomous driving is thus only possible if you have an exact understanding of the overall situation. Whether this information is used for or against a driver, and how this would be done, isn’t a technical decision. There was already plenty of discussion about using data to understand traffic when they introduced the German toll system on gantries over the autobahns.
Dr. Oliver Bühler and Dr. Daniel Ulmer are co-managing directors of Steinbeis Interagierende Systeme GmbH. The Steinbeis Enterprise is a systems supplier and development partner for all kinds of projects related to the testing of embedded systems. Its work revolves around the planning, development, construction, and operation of testing strategies and test platforms for the evaluation of driver assistance systems. In 2015, the Steinbeis experts and Daimler won the Steinbeis Foundation Transfer Award – The Löhn Award for their joint work on the development of an innovative testing environment and development of software instruments for use in advanced driver assistance systems.