Robots to the rescue
September 25, 2012It's the fifth time that engineer Thorsten Lüttel has taken part in the three day European Military Land-Robot (ELROB) competition in Thun, Switzerland. As a researcher, Lüttel focuses on autonomous systems in aerospace at the University of Munich.
But Lüttel and his team did not fly to Thun - instead they have converted a Volkswagen Touareg.
Lüttel's vehicle is packed full of technology. There are camera sensors to scan their surroundings, laser scanners to measure distances around the vehicle, creating a 3D point cloud (a set of vertices in a 3D coordinates system). The data from the sensors is fed into a computer, which uses the data to control the car's throttle, brake, steering and transmission.
But Lüttel's team will have to prove the vehicle can do all that here at the competition.
They have been given a few difficult tasks to complete - such as the mule scenario in which the converted Touareg follows another vehicle.
"The robot doesn't know where it has to go," says Lüttel. "It only knows it has to follow the other vehicle."
Map as you go
In another event, robots are made to find their way around using only the GPS coordinates for their final destination.
Frank Höller from the Fraunhofer Institute for Communication, Information Processing and Ergonomics (FKIE) has entered a small track vehicle in the race.
The small robot has to find its way over rough terrain. To do this, it uses self-generated laser images to create a map. The further it travels, the more detailed the map becomes.
"The robot is just trying to reach its target. If it hits a dead end, it looks at the map to decide where to go next, and it tries that," says Höller.
It's a tough competition.
The organizers set up obstacles like tree trunks, which are hard for the robots to identify.
Based on a previous attempt, Höller says the robots think large puddles look like solid surfaces and they don't recognize soft grass at all.
"The road was elevated and there was grass on both sides, which was exactly the same height as the road," says Höller.
A laser scanner scanned the area, but the computer failed to determine the way forward. As far as the computer was concerned, the whole area was flat.
"So, the robot went into the grass, got bogged down and got stuck."
Searching and finding
In another exercise at the ELROB event, the robots have to find square, orange warning signs, like those found on trucks transporting hazardous materials. They have to photograph the signs and map them.
Computer scientist Torsten Fiolka feels he has risen to the challenge.
He uses his robot laser to both measure distances and the reflectivity of surfaces.
"Warning signs are generally more reflective than other objects," says Fiolka. "So, I use the point cloud of the laser to find highly reflective objects and try to determine whether their dimensions correspond with one of the warning signs."
The event is far from a game. The purpose of the exercise is to develop robots that can be used by the military, the police and fire departments in stressful and difficult situations.
ELROB takes place under the watchful eye of potential customers.
Its organizer, Frank Schneider from FKIE in Wachtberg, says research into robots is more important now than ever before as they become a necessity for emergency services personnel.
"The communication between the system and the user - that is to say the interface between man and machine - is currently being neglected," says Schneider.
People have better sensors
But some emergency services, particularly the fire department, remain cautious about using robots.
They say robots are still too inflexible and slow to be effective in emergencies.
"People are still faster," says Schneider, "and in a disaster situation, every second counts."
Andreas Ciossek agrees. His company, Telerob, builds remote controlled robots for bomb disposal and measuring gases and radioactivity. But he says robots cannot reliably assess whether a building might collapse.
"When you are in there yourself, you can tell if a wall is bulging. You hear cracks, you see fissures. But that's very hard to get with just a video camera," says Ciossek.