When it comes to automated driving, germany is divided into two major camps. Some are looking forward to their first nap in a self-driving company car. The others distrust the machines and don’t want to let go of the steering wheel. And then there is the third group. This consists of the people who are soberly preparing for the future with autonomous cars.
They are currently asking themselves some pretty fundamental questions: how do we actually want to sit in cars that drive without our help?? How much attention will people still have to pay?? And: will accidents have a different effect on us if we don’t even see them coming because we don’t pay attention to traffic at all??
Kristian seidel belongs to this third group of people – and is working to answer such questions. The engineer researches at the institute for motor vehicles, in short ika, RWTH aachen with many colleagues on how carmakers can continue to produce safe vehicles in the future. His project, funded by the EU, is called future occupant safety for crashes in cars or simply OSCCAR. The goal is to improve occupant safety in accidents involving automated vehicles.
autonomous cars will have other accidents
One thing can already be predicted with a high degree of certainty: when automated driving becomes established on our roads, it will not only change the way we drive, but also the way accidents happen. Sounds like a pretty dry research subject, but it’s essential if robot cars are to become a reality.
An example of "new" accidents: when the self-driving car reacts jerkily and even the driver is completely surprised by the movement, high forces can be at work. This is also referred to as pre-crash kinematics, that is at work when the vehicle reacts to an imminent danger. At the moment, this reaction still comes from the driver himself. In doubt, he knows shortly before a crash or full braking what could await him a few milliseconds later and how his body posture and the effect of conventional restraint systems might be affected. What will change when he puts the car in control?.
The research group to which Kristian Seidel belongs does not think in terms of previous structures. "crash dummies can simulate different bodies, but they don’t cover the full spectrum of human physiology," he tells WIRED. That’s why the group relies on a mixture of physical simulation with dummy crashes and virtual crash tests. Thereby come human body models to be used, i.e. digital models of the human being.
The software should achieve an even higher bio-fidelity than tests with plastic dummies. In other words, it should calculate as accurately and realistically as possible how different body types react to the potential loads. Computers should predict which bone fractures and organ damage are to be expected in an accident and how much the occupants will be shaken by the pre-crash kinematics. Changes in demographics are also taken into account. Germany is getting older – and old bones break more easily.
By wolfgang kerler
The autonomy of cars should not yet be overestimated
Human drivers still have to keep a constant eye on the road, because the autonomy level of cars is quite low. "today’s vehicles offer a maximum of level 2, requiring the driver to monitor the system continuously," explains seidel. Level 2 corresponds, for example, to the parking assistant, which works independently up to a certain point – but requires constant supervision and the cooperation of the driver.
The lane assistant of many vehicles also belongs to level 2. This takes the strain off people, especially on long highway journeys. But it doesn’t take away the driving. "Unfortunately, some of the marketing of the systems raises false expectations in this regard," adds the engineer. Drivers would be taking a high risk if they overestimated the autonomy of the systems. "but the goal is level 4 and 5," says seidel. Until carmakers get there, they still have to roll out level 3 nationwide.
In stage 3, the vehicle can be left to run for longer periods of time. It will then change lanes on its own, for example. The driver would only be called upon when decisions are required that go beyond such simple maneuvers on highways. Only at level 4 would the point be reached at which the driver can turn to face his passengers on the often longed-for swivel chair while driving – and start a card game. There is still a long way to go. Even the most optimistic scientific forecasts assume that level 3 coverage of half of all road traffic will not be achieved for at least 30 years. Nevertheless, we should start thinking about what happens if you get hit by a truck while playing a round of poker.
Will we get "seasick" on the highway in the future??
But OSCCAR is not only about potentially bloody crashes, but also about researching our possible user behavior with autonomous cars. The project has only been running for a few months, but the first phase of the study has already been completed. But the results have not yet been published. kristian seidel doesn’t want to venture too far out of the woods – but nevertheless: "getting used to the system could be an important aspect."getting used to cars that steer themselves. Many test subjects still prefer a sitting position facing in the direction of travel, because they have motion sickness feared. In simpler terms: they were afraid of getting seasick. In addition, a leftward gaze was perceived as more pleasant than a rightward gaze. For the project group, however, this could become relative once they get used to the idea.
"the left side is probably more popular because that’s where the traffic is," seidel suspects. But once the sideways sitting position has become the norm, it’s no longer important. "have you ever paid attention to what’s happening in oncoming traffic on a train?? No. They know it’s not necessary to keep an eye on everything outside and concentrate on their laptop or their book."
OSCCAR is about basic knowledge on the one hand, but at the same time it is about real applications. The Aachen-based company already has its own crash facility, where it uses physical dummies in addition to simulations. Research into the safety of automated vehicles is still in its infancy. We all have to be patient, says seidel. "we are still in the learning phase. The systems are now going to driving school, which also takes a while for humans." many scientists are currently working on this topic. Among other things, this is how a database of possible driving and accident situations is built up. A mammoth task: after all, it should include all conceivable critical situations in traffic, so that all possible consequences for the vehicle occupants can also be calculated on this basis.
Automakers support the project
According to seidel, the ultimate goal must be for the capabilities of automated vehicles to be at least equal to those of an average human driver. The three-year OSCCAR project is not expected to produce a market-ready solution in the end. Nevertheless, it can be assumed that the results could have an influence on the architecture of future cars. Because various carmakers are on board with the project. Whether poker will be played when driving in 30 years is still an open question. But it seems quite possible. We should therefore be prepared for.