Automation Scientists Fredrik Danielsson and Bo Svensson are developing flexible automation systems (Photo : University West)
The very nature of robot programming makes them susceptible to the eventual wrench in the cogs. One misstep and the entire operation falls apart - which is why researchers from University West in Sweden have figured out a way to allow robots to think for themselves and function locally, even when things go wrong.
Robots are a highly standardized practice in the manufacturing industry. Everything from canned goods to cars can be put together using a complex series of machines that carry out assembly line tasks. If one of these machines hits a snag, however, the entire process comes to a grinding halt thanks to the hierarchical nature of the production line's programming. This is the problem automation scientists Fredrik Danielsson and Bo Svensson set out to solve.
"A single error somewhere makes everything stop. For example, if a sheet metal is damaged an operator has so take it out and then reset and restart everything," says Svensson.
What Danielsson and Svensson were able to do is create a production line where the entire assembly line doesn't shut down just because one component hit a snag. They demonstrated this with a production line consisting of three robots, two metal cutting machines, a transportation system, a material handling system and a measuring station. Rather than every single part playing slave to a master control system, each component in their production line was equipped with an 'agent,' a unique intelligent program that allows each piece to operate for itself.
"The agents know what neighbors they should communicate with and make small local decisions," explains Danielsson.
These robots don't make decisions like which pair of shoes goes with a certain outfit or whether or not to splurge on that fancy mink coat. Their decision making ability means they act based on local criteria and circumstance, not on some hardwired predetermined series of steps.
Each robot in this production line acts according to tasks being performed nearby. This means that a robot 20 steps into production won't stop working because there's a holdup in step 3. Instead, this production line accepts inputs in the middle (such as replacing a damaged sheet of metal) without disrupting the entire production process.
This kind of dynamic thinking has its obvious applications in the industrial sector. The agents are easily reprogrammable, meaning a variety of tasks can be given to the same set of machinery. In essence, Danielsson's and Svensson's production line can retain the same efficiency of a motorized assembly line, while gaining some human flexibility. While the system is currently expensive to implement, with further advances, this can only spell out more cost-efficient manufacturing processes that will mean greater variety and affordability for the consumer.
Outside of the industrial sector this kind of system programming may one day find it's way into simple software circuitry. In computers, for instance, a problem with a portion of the hardware that isn't entirely necessary for the current task doesn't mean the entire system has to crash. By allowing each component to adapt itself to what is available and required, this could allow for more stable electronics. From there, it's not hard to imagine implementing this in any hierarchical system with a master control in place, such as electrical grids.
Check back tomorrow for another installment of our Applied Robotics series titled: "Applied Robotics: How Robots Could Change Life - A Real-Life Avatar Program."