MIT engineers build, test bartending robots that work together

BOSTON, Aug. 12 (UPI) -- Engineers at MIT have designed a team of robots capable of working together to pour and deliver beers to thirsty humans.

Researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) have essentially mechanized an entire bar waitstaff. One robot cracks open beers, while two others take orders from patrons and deliver the suds.

A special PR2 robot served as bartender while two four-wheeled Turtlebot robots operated as the wait staff.

What's notable about the MIT lab-turned-watering hole (and its robotic workers) are not the individual skills on display, but the teamwork. Engineers say advancements in robotics communication are essential for the use of robotic teams in dynamic and chaotic real-world human scenarios, where new information must be processed in real time.

MIT's Turtlebots are programmed to anticipate what drinks are needed where, taking orders and delivering drinks with the greatest possible level of efficiency.

The robots are governed by advanced algorithms that decline to micromanage. Instead, the robots' software allows the team to understand the general problem, leaving its programmed intuition to sort out the specifics of the best problem-solving strategy.

Beer delivery is no small matter, but there are admittedly more pressing issues out in the real world. Ultimately, researchers want to create robots that can assist in unpredictable, high-pressure environments -- like hospitals or search-and-rescue scenarios.

Researchers purposefully program knowledge gaps into the robot team, in order to simulate the disorderly nature of the outside world.

By working together with state-of-the-art communications systems and a more hands-off programatic approach, the robots are better able manage uncertainty.

"Each robot's sensors get less-than-perfect information about the location and status of both themselves and the things around them," MIT graduate student Ariel Anders said in a press release. "As for outcomes, a robot may drop items when trying to pick them up or take longer than expected to navigate. And, on top of that, robots often are not able to communicate with one another, either because of communication noise or because they are out of range."

"These limitations mean that the robots don't know what the other robots are doing or what the other orders are," Anders explained. "It forced us to work on more complex planning algorithms that allow the robots to engage in higher-level reasoning about their location, status, and behavior."

Previously employed algorithms have proven too complex to scale up to real-world scenarios and problems. The latest software attempts to simplify robotics code by absorbing unpredictability into the algorithms.

"Almost all real-world problems have some form of uncertainty baked into them," said Chris Amato, a former CSAIL postdoc researcher, now a professor at the University of New Hampshire. "As a result, there is a huge range of areas where these planning approaches could be of help."

Amato is the lead author of a new paper on the macro-approach robotics software, published in the journal Robotics Proceedings.