A growing number of manufacturers are running into the same wall. Traditional robots are excellent at repeating the same motion over and over, but they struggle the moment a task changes even slightly, whether that means picking up an oddly shaped part or packing a soft, irregular item into a box. Solving that problem sits at the center of a new partnership between Grid Dynamics Holdings, Inc. (NASDAQ: GDYN), an AI transformation company based in Silicon Valley, and Doosan Robotics Inc. (KRX: 454910), a South Korean maker of collaborative robots, or cobots, used in 45 countries around the world.
The partnership centers on Grid Dynamics’ GAIN Platform for Physical AI, a software tool built to help industrial users design robotic manipulation workflows, roll out the latest physical AI models, and monitor robotic production lines through digital twins, which are virtual replicas of physical equipment. Paired with Doosan’s cobots and Grid Dynamics’ integration services, the goal is to give factory and warehouse operators a fuller package rather than separate pieces of hardware and software they would otherwise have to combine on their own.
That combination is meant to tackle jobs that have historically been difficult for automation. Inspecting objects with complicated or irregular shapes, assembling products where individual components vary slightly, and packaging items that are soft, unevenly shaped, or simply unfamiliar to the system are all tasks conventional robotics software has struggled to handle consistently. Physical AI, which blends machine learning with real world sensing and movement, is designed to close that gap by letting robots interpret what they encounter rather than follow a fixed script.
Beyond the technology itself, the two companies say they intend to build a joint go to market strategy so the combined hardware and software offering reaches industrial customers more directly, instead of through separate sales channels. They also plan to keep researching what is known as robotics policy control, a term referring to the decision making logic robots use to choose how to act in a given situation.
Ilya Katsov, Grid Dynamics’ chief technology officer for the Americas, said the company’s aim is to give industrial users production ready tools for high value tasks such as dual arm assembly and packaging. Michael Ryu, general manager and head of Doosan Robotics Europe, described the collaboration as a way to combine Doosan’s robotics with Grid Dynamics’ AI expertise and global reach, making it easier for cobot users to adopt newer automation technology. Grid Dynamics, founded in 2006, has built its business helping large enterprises put AI to practical use, with offices spanning the Americas, Europe, and India. Doosan Robotics, established in 2015 and listed on South Korea’s KOSPI exchange since October 2023, has built a reputation for cobots that pair 3D vision with what it calls perception-based learning, including a system called Scan and Go that automates tasks such as sanding and grinding on curved surfaces.
Physical AI is still an emerging category, and its real value will depend on whether it proves itself in working factories and warehouses rather than in demonstrations. Still, a robotics manufacturer and an AI focused software company choosing to combine forces this directly suggests both see practical automation problems, not just new algorithms, as the real test this technology has to pass.
