Safety for Dummies

Road transport safety is a top priority across the board—for public authorities, vehicle manufacturers, and citizen drivers alike. In many cases, improving safety means reducing the impact of human error.

Advanced technologies such as Autonomous Emergency Braking (AEB) systems can now take control of a vehicle in situations where a collision, with another vehicle, a roadside object, or a pedestrian, is likely. But a real concern, especially with autonomous systems, is whether the systems themselves are actually safe.

Needless to say, full testing is crucial before any new and advanced technology is ever deployed in a real vehicle on a real road. And that’s where 4activeSystems GmbH comes in.

The Traboch, Austria–based company’s General Manager Martin Fritz says, “We work in the field of active vehicle safety with our main focus on facility technologies and dummies, which we use to test driver assistant systems, such as AEB. Because we use GNSS instead of light barriers for controlling our system, we have improved performance in crash-point precision. Controlled scenarios we could never have imagined are possible now with GNSS in our system.”

The Test Subject

fig1An AEB system comprises a set of sensors to monitor the environment surrounding a vehicle, to the front, side, and rear. Inside the vehicle, sophisticated software works to recognize situations where relative speed and distance between the critical objects and a moving vehicle suggest a collision may be imminent.

“Once the sensor detects an object,” Fritz explains, “algorithms may classify it as relevant, for example, as a pedestrian or bicyclist. After positive classification of a particular object, the system starts tracking it, developing a prognosis of its path.” If this path intersects with the vehicle trajectory, emergency braking can be automatically applied, once the driver is warned, to avoid the collision or at least to minimize the effect.
“The main actor in our test systems is a road-crossing, human-like dummy, controlled by a special drive unit,” Fritz says. “Essentially, the dummy is meant to get in the way of an oncoming vehicle, with a projected impact to occur with a specific part of the front of the vehicle.”

4activeSystems dummies—which, by the way, are sometimes hit by a vehicle during testing—simulate adult and child pedestrians and cyclists. However, they are recognized as real humans by the vehicle’s mono and stereo visual camera systems, as well as radar and infrared systems. In tests, the dummies elicit a homogeneous distribution of the Radar Cross Section (RCS)—a measure of the detectability of an object by radar—with the RCS values remaining relatively constant from different views. For infrared sensors, the dummy is 50% reflective in the spectrum between 850 and 950 nanometres.