Automation has helped manufacturers increase productivity and lower costs. Yet, automation’s potential has barely been tapped. Robots today perform only 10% of tasks in a plant, on average (source: Boston Consulting Group).
What of the other 90%? Many of these tasks had, in the past, proven too costly to automate due to frequent line changeovers, lower production runs, and the extensive fixturing and workcell redesign required to perform the tasks.
Until five years ago, the idea that robots could perform the skilled tasks requiring a “human touch” seemed impossible. No way could a fixed traditional robot delicately place a thin circuit board on a tester and then remove it – all without breaking it or the tester or harming someone in the process.
Today, Rethink Robotics’ smart collaborative robots (cobots) are doing this task and many others – including machine tending, material handling, and inspection – all the while working side-by-side with their human co-workers.
How did Rethink engineers design robots that work the way people do?
The answer is actually pretty low tech: we went into hundreds of factories armed with notebooks, measuring tapes, and stopwatches and watched people work.
Our assignment: catalog tasks
Before even thinking about how to design a new type of collaborative robot, Rethink engineers first had to understand the myriad tasks performed on the factory floor and how people perform them.
- Did the person need to stand, sit or both? How long did each action take and when did it occur in the process?
- Did the person need to bend, turn or extend the arm or upper body while performing the task?
- How many actions did the task require and how long to perform each one?
- Did the worker need to accommodate changes in the line movement e.g. slowing down, speeding up, a crooked box, etc.?
- Did the person have to go walk somewhere as part of the task and why?
Once we collected the data, we then grouped similar tasks and analyzed them. Our goal was to create a new type of collaborative robot that could address the largest number of tasks that formerly couldn’t be automated.
Designing for reach
Human reach translated to a robotic arm is particularly important. Think of the analogy of baking cookies. You measure, sift, stir. You crack an egg. You may stand at your kitchen counter working, but at some point you have to turn, reach up for a bowl on the shelf or bend and twist for the baking sheets in a cabinet.
When designing Sawyer, we accommodated for these types of movements – which is why the robot has one arm with seven degrees of freedom (DOF) and approximately four feet of reach – about what a person needs to extend the arm and body to reach for something.
When cataloging tasks, we also had to look at how people use their hands. Did the worker need to twist the wrist or pinch the fingers? Did the task require the use of both hands or only one and which of the hand actions were repetitive?
As a result of this “go look, go see” work, Rethink engineers were able to design smart agile robots that mimic how people work – right down to having the ability to delicately place circuit boards onto testers without breakage.
To see examples of our cobots at work on the factory floor, visit the Rethink Robotics video gallery. And to get a grip on the rest of the Rethink Robotics blog, subscribe above to receive blog post alerts delivered directly to your inbox.
About the Author
Jim had a choice upon graduating from Tufts University – chase a dream as a concert pianist or become part of the inaugural Leaders for Manufacturing Program at MIT. He chose the latter– dedicating his career to developing and delivering innovative solutions that improve the business of manufacturing. Internally at HP, and then at breakthrough start-ups in e-commerce, inventory optimization and supply chain risk management, Jim’s never once looked back. His charter today: capture the power of data and analytics to push the standard for world-class manufacturing higher—once again.