Crossroads 2014: Dances with Robots

Machine age. Robots, such as this unit from Universal Robots,  can now work alongside humans on the factory floor. Photo: Universal Robots
Machine age. Robots can now work alongside humans on the factory floor. Photo: Universal Robots

Robots have been an integral part of the industrial landscape for many years. Advances in the technology are now making these units more flexible and versatile. The upshot from a supply chain perspective is that the new generation of factory robots makes production systems more agile and helps companies to compete in fast-changing markets.

A key development is in “scheduling the choreography of robots,” explains Dr. Julie Shah, Head of the Interactive Robotics Group, MIT Computer Science and Artificial Intelligence Lab. Shah will explain the implications at the forthcoming MIT Center for Transportation & Logistics’ Crossroads 2014 conference, March 25, 2014, on the MIT campus, Cambridge, MA.

The movement of humans and robots on the factory floor has to be carefully choreographed. A certain distance between people and machines has to be maintained for both safety and efficiency reasons. Another constraint is that work assignments must be completed within a predetermined time; between paint applications or before the end of a shift, for example. Some tasks can only be carried out in specific locations.

Moreover, these requirements have to be met without disrupting work flows. “People must be able to move in and out of a space without the need to shut down a cell (a team of robots),” explains Shah. Similarly, hindering machine movements because of the proximity of human operators is to be avoided.

Constraints like these are not new. However, robots are now being deployed in areas that were once the exclusive preserve of humans, raising new challenges for manufacturers.

“In the past, these have not been major issues, especially when there are structured tasks for robots, but increasingly robots are being used to perform work that used to be carried out by people,” says Shah.

Also, advances in robot technology mean that many machines no longer have to be “caged” behind special protective barriers, opening up new possibilities for integrating these units into production processes.

The end result is more complex interactions between humans, machines, and the manufacturing processes that ultimately feed markets. If this “dance” can be reorganized on the fly to accommodate, say, an unexpected production changeover, then the entire operation becomes much more responsive.

“You can quickly optimize work scheduling if, say, different batch sizes are introduced,” says Shah. It is also easier to eliminate potential production delays when work schedules can be reconfigured in near real time.

It’s not difficult to imagine the supply chain implications. The explosion in online sales and the proliferation of SKUs owing to product customization are just two trends that are driving increased demand volatility. Improving the ability of manufacturing systems to pivot with shifts in demand effectively makes the end-to-end supply chain more agile.

This is what a team of researchers, including Shah, are working towards. They have developed an algorithm called Tercio that can reconfigure a work schedule involving multiple agents very quickly. For example, using traditional methods, it can typically take 30 minutes to reschedule four robots assigned with some 20 tasks. Using the new algorithm, 10 robots responsible for 500 tasks can be reassigned in 10 seconds or less in response to an external change.

“It can take a few hours to compute a new schedule even when minor problems occur, but our algorithm is the first to handle scheduling problems like these in seconds,” says Shah.

Tercio will also open up new functional horizons for robots, she maintains, reinforcing the trend towards the wider use of these machines in production systems.

The team is working with manufacturers in a number of industries to apply the algorithm in real-world production systems,” One example “is to dynamically schedule assembly operations so that the workflow is near-optimal despite variable process times,” says Shah.

Register for the Crossroads 2104, Biomanufacturing, Robots, and 4D Printing: The Next Decade of Disruptive Innovation, conference here.

This article appears in the Winter 2014 issue of Supply Chain Frontiers, the newsletter of the MIT Global SCALE Network. Subscribe to Frontiers for free here.


Leave a reply