For many manufacturers, the era of mass production is drawing to a close. Changes in consumer behaviour have forced them to rethink their ‘economies of scale’ approaches as they reorganise to answer the demand for more personalised goods and services. This customisation means items have to be produced on demand, rather than sold from stock, and the modern customer is fussy and impatient, so it has to be done fast and with no compromise in quality.
In the automotive industry, for instance, models are becoming available with an ever larger variety in the number of possible feature combinations. This presents a problem for automated robotic assembly lines because each change requires a time-consuming reconfiguration of the equipment. The current limitation of even the most sophisticated artificial intelligence (AI) algorithms is that they are designed to perform one task – and one task only. The software that drives an autonomous vehicle is incapable of playing a game of chess.
In response to this, researchers at Siemens are investigating how robots can teach themselves to perform new tasks. Based on a promising new AI technology called deep-learning, the method makes use of CAD files containing information about desired colour schemes, geometry, final assembly, choice of finish, and the like.
In simplistic terms, the AI algorithms embedded in the robot interpret the various CAD models to generate the appropriate programming instructions in response to a new production order. The robot itself decides the sequence in which tasks should be performed, and also corrects faults as and when they occur during the assembly process. Manufacturing’s Nirvana – aka batch size one – seems almost within reach.
Once these deep-learning techniques are perfected for industrial use, AI has the potential to transform manufacturing much as electricity did some hundred years before it. Until then though, artificial intelligence’s contribution to the industry will likely remain confined to data mining applications in areas like energy efficiency, quality control, condition monitoring and predictive maintenance. This does not mean these systems are not powerful in their own right, just that they are not ready to take over the world quite yet.
Industry guide
Speaking of predictive maintenance, posted with the magazine this month is the 2019 edition of the Technews Industry Guide: Maintenance, Reliability & Asset Optimisation. This one-stop resource for the modern maintenance professional covers everything from in-situ sensor-based solutions for condition monitoring, through handheld portable devices for periodic maintenance-related checks, through software solutions for analysis and reporting, and on to customised services like reliability management consulting and training. Our hope is that the ideas and insight we have gathered together will help you to solve a problem you may be faced with in your own particular plant.
Steven Meyer
Editor: SA Instrumentation & Control
| Tel: | +27 11 543 5800 |
| Email: | malckey@technews.co.za |
| www: | www.technews.co.za |
| Articles: | More information and articles about Technews Publishing (SA Instrumentation & Control) |
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version