"Navigating an airport is challenging, there is a lot of glass and a constantly changing environment in terms of temporary obstructions, such as parked luggage trolleys and people everywhere," Achim Lilienthal, project leader at Örebro University which contributed to the research, said in a statement. "Objects that are temporarily permanent so to speak, are the most difficult to work around. We do not know, for instance, how long that luggage trolley will be parked in a particular spot, which makes it harder for the robot to determine its own location. We are working on a general map representation that includes and allows the robot to handle temporarily permanent objects."
The team has also trained Spencer to understand human behaviors and, to some degree, social mores. For example, the robot will not simply barrel through a crowd of people, but rather sidestep the group. It's also cognizant enough to occasionally check on the group it's leading to ensure that nobody's been left behind. "This technology can be used in all robots intended to interact with humans. Autonomous trucks for example, would be more widely accepted if they functioned better in their interaction with humans," said Lilienthal.
The first trial run begins November 30th and will last just one week. The research team will then leverage the data they collected from the trial to further improve Spencer's performance before its official debut next March.
[Image Credit: LASS Laboratoire]