Michael Harper for redOrbit.com – Your Universe Online
German scientists are looking to improve human to robotic interaction by setting the two up together at the bar.
Modern science has already produced more than one robotic bartender capable of pouring spirits into cups, giving them a shake and serving them up to thirsty bar patrons. These bartenders have to be told what to do via smartphone app, however, and cannot anticipate when someone wants a refill.
Professor Jan de Ruiter with Bielefeld University in Germany has been working to improve robotic technology to the point where the machines can recognize visual cues and body language — but first he had to understand it himself. He and his team of researchers studied the way people order drinks at the bar and found the most effective way is also the simplest — walking directly to the bar, looking square at the bartender, and asking for a drink.
Dr. Ruiter’s research is now published in the journal Frontiers in Psychology.
What the research team has learned is being programmed into a robotic bartender called James, or Joint Action in Multimodal Embodied Systems. The apron-less robot has a spartan figure with a torso that spins on a rotator and a single arm with a hand for grasping drinks. Dr. Ruiter’s team use an iPad for James’ head, and a pair of large cartoonish eyes and a thin smile make up his face on the tablet screen.
As customers approach James’ bar, cameras can recognize their facial expressions as microphones listen to the drink order. James then goes to work, reaching for the drink with a four-fingered hand. In a video James is only seen grabbing plastic bottles and doesn’t actually mix any cocktails or pull a beer from the tap.
“In order to respond appropriately to its customers the robot must be able to recognize human social behavior,” said Dr. Ruiter in a statement. “Currently, we are working on the robot’s ability to recognize when a customer is bidding for its attention. Thus, we have studied the process of ordering a drink in real life.”
Without a clear direction and definition of facial cues and body language, James might never pour a beer for a patron.
To understand how people order drinks, Dr. Ruiter’s team set up video cameras in clubs and pubs in Germany and the United Kingdom and analyzed the way people ordered. They found few people actually signal a bartender by fidgeting with or looking at their wallet. Even fewer people were served when they were looking at a menu or drink list. The most effective way to get a bartender’s attention and be served a drink, say the researchers, is the direct approach.
“Two signals are necessary and together form the sufficient set of signals for identifying the intention to place an order,” Dr. Ruiter told the Telegraph.
“First, the customers position themselves directly at the bar and, secondly, look at the bar/bartender. If one of these signals was absent, the participants judged the customers as not bidding for attention. This provided a clear indication that both signals are necessary for bidding for attention.”
After analyzing their research, the scientists found 95 percent of all customers who received their drinks within 35 seconds directly confronted the bartender in this way.
Dr. Ruiter and team have now programmed James with this information and say the robot can understand when a customer is ready to place an order. James has also been programmed to work on a first-come-first-served basis and will ignore anyone who cuts in line.
The researchers have been working on James since early 2011 and hope to have the project completed in January 2014.