Industrial robots will learn & adapt on the fly, very soon

Industrial robots will learn & adapt on the fly, very soon

Industrial robots will learn & adapt on the fly, very soon

When you think of future factories, imagine robots that can interpret human instructions in everyday language, figure out their own work plans, and master new tasks by studying videos – without being painstakingly programmed line by line. This vision is rapidly becoming reality, says Marc Segura, president of ABB’s robotics division.
The Swedish-Swiss multinational behemoth has installed close to half a million industrial robots worldwide and its R&D hubs – particularly in India – are harnessing advanced AI and generative models to push robotics far beyond its traditional limits.
“Generative AI is a game-changer,” Segura told us recently on a visit to India. “We’ve spent years on deep learning, but in just a few months with generative AI, we’re seeing results that surpass what we had after six years of previous work.”
At the heart of this transformation lies the integration of AI-based perception, navigation, and planning capabilities into ABB’s robotic platforms. Gone are the days when robots only excelled in highly structured environments. Now, they can recognise and pick unknown objects, navigate complex factory floors using 3D mapping, and even adjust their movements on the fly.
“We’re going to be talking to robots in natural language,” Segura says, “and they’ll translate those instructions into actions, developing increasingly dexterous skills.”

Marc Segura

Early AI integration
ABB’s journey with AI in robotics started more than seven years ago. Initially, deep learning neural networks were used to solve unstructured environment problems – like item picking. The robot was taught to take a picture of a bin of mixed objects and segment that image into distinct items, and then guiding the robot to pick each one. Over time, ABB refined these solutions to achieve reliability levels essential for industrial applications. “When you go to a large e-commerce or parcel company, they demand near 99% performance,” Segura notes. “We had to industrialise AI-based picking to reach that.”
One of the key AI applications is in machine navigation. “We use 3D cameras and visual SLAM (simultaneous localisation and mapping) technology in our mobile robots,” Segura explains. “They understand what’s around them – the difference between a human and a column, for instance. A human might step aside, a wall won’t. That semantic understanding helps guide the robot’s collision avoidance and path planning.” The result: robots that not only know where they are but can also react intelligently to who or what is in their way.
Generative AI took this evolution a step further. “We invested years in refining certain capabilities,” says Segura, “and then, with generative AI, we enhanced them within months.”
Path to true autonomy
Today’s advancements hint at a future where robots could become truly autonomous. While fully independent robot workers remain a long-term goal – Segura estimates five to ten years out – progress is constant. “We’re seeing a move toward foundational models for robotics, similar to those in language AI,” he says. “By combining force feedback, visual data, and large data sets of human demonstrations, we’ll get robots to do more complex, dexterous tasks in unstructured environments.”
Rather than being confined to static assembly lines, future robots will pick up a wide variety of skills. The ultimate aim is a system that can respond to unpredictable scenarios – imagine a single robot line worker that can adapt to new components or product variations on the fly. “We believe a fully autonomous robot that adapts like a human is coming,” says Segura. “But it will take time and incremental breakthroughs.”
India’s R&D role
While mature markets show robotics growth leveling off, India is racing ahead. “India is a highlight in the world right now,” Segura says. “Globally, the robotics market has been flat, but India hit its highest growth last year, becoming the seventh largest robotics market worldwide.” Despite this, India’s robot density – robots per 10,000 manufacturing employees – is still around 148, far behind the 500-plus in advanced economies. “There’s huge potential,” he says, noting that with such a large runway for growth, India could leapfrog directly to advanced robotics solutions.
For ABB, India’s significance goes beyond being a growth market; it’s a critical innovation hub. “India is a powerhouse for us in terms of competence,” Segura says. “We have an R&D hub here with engineers designing robots for the entire world, and an engineering operations team supporting global efforts.” The Indian teams work on software like Robot Studio, ABB’s digital twin simulation platform introduced in the late 1990s and now transformed into a cloud-based tool. They also contribute to applications like Pick-Master, ensuring that the software backbone of ABB’s robotic solutions remains at the cutting edge.
India’s talent base allows ABB to accelerate innovation cycles. “Our team in India collaborates closely with Sweden and China,” Segura says. “They’re working on software and controls that will make robots easier to deploy and adapt, which is essential for helping manufacturers move quickly in a volatile market.”


link

Leave a Reply

Your email address will not be published. Required fields are marked *