Releasing robotic assistants from their cages

Could a new algorithm make it safe for humans and robots to work side by side?
Written by Charlie Osborne, Contributing Writer on

Could a new algorithm make it safe for humans and robots to work side by side in manufacturing plants?

It seems so. Researchers at MIT have been pondering the question -- particularly, if it would ever be possible to release the locks on cages that protect human workers from their synthetic counterparts in the manufacturing industry.

Currently, the divide between workers is straight and clear. Robotic workers focus on heavy duty repetitive tasks, whereas humans are assigned roles that are less hazardous and require a higher level of finesse.

However, this may not always be set in stone. Julie Shah, the Boeing Career Development Assistant Professor of Aeronautics and Astronautics at MIT, believes that the future of factories could mean hosting humans and robots who work together -- especially in manufacturing specializations like airplane construction.

The idea is simple: If robotic helpers can be made safe enough to allow them to perform tasks to improve human efficiency, together, then it could potentially revolutionize manufacturing. Shah said:

"If the robot can provide tools and materials so the person doesn’t have to walk over to pick up parts and walk back to the plane, you can significantly reduce the idle time of the person.

It's really hard to make robots do careful refinishing tasks that people do really well. But providing robotic assistants to do the non-value-added work can actually increase the productivity of the overall factory."

Currently, how to perform tasks repetitive tasks is simply programmed into a robotic assistant. If you wanted to make them useful to an individual, the technology would have to be able to adapt to each mechanic -- and this would force robotic design to a new level.

Shah and her team at MIT have been researching this idea, and now have developed an algorithm that potentially allows a robot to quickly learn and process individual preferences for particular tasks, and then adapt their own modes of behavior accordingly. The findings will be presented at the Robotics: Science and Systems Conference in Sydney in July.

"It's an interesting machine-learning human-factors problem," Shah said. "Using this algorithm, we can significantly improve the robot's understanding of what the person’s next likely actions are."

An example Shah's team used to create the algorithm is that of spar assembly -- the way you construct an aircraft wing. Where one human worker may prefer to apply sealant to predrilled holes before hammering in bolts to the wing en masse, another may complete a line of bolts one at a time.

So, how would a robot 'know' how best to help each individual mechanic? The computational model used was constructed in the form of a decision tree -- set with different options that a mechanic may take.

Then, the robot observes, learns, and takes a particular route down the decision pathways. However, the robot still needs to be able to identify different mechanics in order to choose the best option.

By using radio-frequency identification (RFID) tags, individual workers could be recognized by their robotic counterparts. Once a single robot has learnt a person's individual habits, this data can be made available to the rest of the synthetic team.

Steve Derby, associate professor and co-director of the Flexible Manufacturing Center at Rensselaer Polytechnic Institute, believes the researcher's efforts may bring true, intuitive collaboration between robots and humans a little closer.

Derby said:

"The evolution of the robot itself has been way too slow on all fronts, whether on mechanical design, controls or programming interface. I think this paper is important -- it fits in with the whole spectrum of things that need to happen in getting people and robots to work next to each other."

This research was supported in part by Boeing Research and Technology and took place collaboratively with ABB.

(via MIT)


This post was originally published on Smartplanet.com

Editorial standards