MT Magazine May/June 2026

15

FEATURE STORY

MAY/JUNE 2026

Talla explained: “If you look at robotics and automation for the last 50 years or so, it’s been deployed mostly in the high-volume, low-mix applications – things like the automotive industry and high volume electronics. But much of the world is in small and medium enterprises.” He said that a problem is that setting up and programming the robots can be too expensive for the smaller operations, particularly because they tend to have a high mix of tasks rather than the repetitive production characteristic of larger operations. The solution? “The only way to solve that is to use a reasonably good enough general-purpose AI brain, and that’s what we’re able to do with the latest-generation AI technology,” Talla said, adding, “And before you actually go and deploy this in the factory or your manufacturing facility, you want to be able to test it in simulation because it’s faster, safer, and cheaper.” Can It Be Done? All of that said, simulation, AI, and the like may appear beyond the competence of small or medium-sized manufacturers. Foxconn using this tech is one thing – but it’s another for less-capitalized shops. So, MT Magazine asked whether shops would need to find hires from Stanford University or the Massachusetts Institute of Technology to use this tech. Talla answered with an analogy – the release of ChatGPT in November 2022: “Two fundamental things happened, right? The first one, of course, is that the model was general purpose – that it could do multiple tasks. “Before ChatGPT, you had AIs for specialists in one specific task. ChatGPT changed that with being a good generalist model. The second thing was, of course, it was so easy to use for anybody literally on the planet. That’s the same idea we have for robotics experience.” He said that the object will be to go from traditional robot programming to teaching or instructing them what to do – and doing so in a way that will be accurate in the physical world. Do You Need a Humanoid? Another development in the automation world is the introduction of humanoid robots. Although Tesla’s Optimus may be the most famous, there are several, including the Figure series from Figure AI, Apollo from Apptronik, and Digit from Agility Robotics. FANUC is among the largest industrial robot companies in the world. Mike Cicco is president and CEO of FANUC America Corp. Although the company doesn’t offer bipedal, human shaped robots, Cicco said that if the need is for two arms working together, they can provide that – in fact, one control can orchestrate four arms. If the situation requires a robot that can move on the factory floor, they have that, too: mount a robot (or two) on a mobile robot, and it can go where it is needed.

microservices, and real-time 3D technologies that can be used to robustly and accurately create and simulate industrial twins. AMT – The Association For Manufacturing Technology is working with Nvidia on Omniverse implementation in manufacturing operations, including deployment with MTConnect. Deepu Talla, Nvidia vice president of robotics and edge AI, said, “The industrial sector needs physically accurate simulation to bridge the gap between virtual training and the real-world deployment of AI-driven robotics at scale.” If you’re going to create a simulation for training robots to perform tasks, making it as close to physical reality as possible is important. ABB claims manufacturers will see significant benefits by using RobotStudio HyperReality to create production lines: • Setup and commissioning times can be reduced by up to 80% • Costs can be reduced by up to 40% (because there is said to be such fidelity with the physical world, physical prototypes aren’t necessary) • Time-to-market for complex products can be reduced by 50%

Working with Nvidia, ABB Robotics is creating more accurate and realistic simulations for training robots. The scene above is what could be created with its simulation system; below is the Nvidia-powered version. (Image courtesy of ABB)