Pharma, Robotics and Jidoka: Perspectives from ABB’s Bengt Stom

As we found in developing articles for our August issue, pharma is embracing the Toyota Production System (TPS), but mainly one pillar of that system: Kan ban, or Just in Time.  Waiting to be developed is Jidoka, the other pillar of TPS.  People can't always agree on what it means says U.K. based ABB Robotics pharma industry segment manager, Bengt Stom.   We'd asked Bengt for his views on Jidoka and for pharma's readiness to embrace robotics.  Here's some of what he had to say. PM Why isn't use of robotics gaining traction faster in pharma? B.S. That pharmaceuticals companies have been reticent in addressing the use of robotics stems largely from the recent history of producing very large volumes of similar product 24/7 with no need for flexibility. Such production often lends itself to fixed automation whereby recipes never change. However, as pharma companies [move away from the blockbuster drug model], the need for  flexibility in both volumes and packaging types will pressure manufacturers to [look for new ways to] introduce ever greater flexibility.This is where robotics technology always beats fixed automation systems.Of course, there is also the perfectly normal human prejudice of ignoring the unusual or unfamiliar because it challenges conventional thinking and experience. PM How is Jidoka being incorporated into robotics products for pharma today? B.S. There is widespread confusion over how to define Jidoka.Some interpret this as simply the ability to automatically stop a system if things go wrong. I prefer the more holistic concept that the automation system is intelligent enough to assess what is going on and act accordingly - rather as a human might. What is needed is a combination of software and hardware.  By incorporating vision systems, weighing transducers and proximity sensors, the robots can be instructed to act in accordance with the performance of the overall system.  In other words, if a line slows, or product flow falters, the robots simply adjust their actions accordingly.  The key to the integration of this proven hardware lies in the ability of the software to integrate the robot controls with the senor data provided.  This software is also made simple to use in order to open up the possibility for increased machine intelligence to everybody. PM How far is the industry from embracing the concept of artificial intelligence BS Again, the question begged is "what constitutes artificial intelligence?". If this simply infers the ability of a machine to make rudimentary "human like" assumptions and decisions, then many automation systems already have artificial intelligence.  However, if what is meant is the ability of machines to learn, then I think this is some way off yet.  The solution to true artificial intelligence lies with computer scientists, ancillary sensor designers and software developers