Grid Dynamics and Doosan Robotics Partner on Physical AI Stack

Grid Dynamics has formed a strategic partnership with Doosan Robotics to deliver an integrated physical AI solution for industrial automation, combining collaborative robot hardware with advanced AI software for manufacturing and logistics applications.

Under the agreement, Grid Dynamics will provide Doosan's cobot customers with its GAIN Platform for Physical AI, an end-to-end software stack that handles robotic manipulation, digital twin simulation, computer vision, and IoT monitoring. The partnership aims to enable automation of tasks that have proven difficult for conventional robotics systems, according to details first reported by Interesting Engineering.

What the platform delivers

The GAIN Platform uses Vision-Language-Action models to execute AI-driven tasks including pick-and-place operations, flexible assembly, and bimanual workflows on ROS-compatible robots. Digital twin capabilities powered by NVIDIA Omniverse allow manufacturers to simulate and validate production lines before physical deployment.

The software also monitors production processes in real time using vision-language models to detect safety violations, verify packaging accuracy, and track workflow performance. An integrated IoT control tower aggregates data from robots, machines, cameras, and sensors to support predictive maintenance and anomaly detection.

Grid Dynamics reports the platform has achieved inspection cycles up to 1,000 times faster and anomaly detection 24 times quicker in existing industrial deployments.

Doosan's collaborative robot portfolio

Doosan Robotics manufactures collaborative robots designed to work safely alongside human workers using integrated force and torque sensors. Unlike traditional industrial robots that require safety cages, these cobots detect collisions and limit contact forces.

The company's product line includes the lightweight A Series for general automation, the M Series for high-precision work, the H Series for heavy-duty applications with payloads up to 25 kilograms, and the P Series with NSF-certified food-grade models. Doosan cobots currently operate across manufacturing, logistics, food and beverage, electronics, automotive, and healthcare sectors.

Target applications

The combined solution addresses automation challenges involving complex-shaped object inspection, assembly with variable components, and packaging of previously unseen or deformable items. The companies plan to pursue a joint go-to-market strategy delivering the hardware-software stack as a turnkey solution.

Beyond commercial deployment, Grid Dynamics and Doosan Robotics will collaborate on research and development focused on robotics policy control technologies that govern how robots learn and execute complex manipulation tasks.

"Our mission is to provide industrial users with production-grade solutions for unlocking high-value use cases such as dual-arm assembly and packaging," said Ilya Katsov, CTO Americas for Grid Dynamics. "Our partnership with Doosan enables the full-stack solution approach and more efficient go-to-market execution using global footprints of both companies."

Why it matters

Manufacturers face persistent challenges automating tasks that require visual judgment, handling variability, or adapting to changing conditions. Physical AI represents a shift from rule-based programming to systems that can generalize from training data, potentially expanding the economic viability of automation to operations currently too complex or variable for conventional robotics. The partnership combines established cobot hardware with emerging AI capabilities, offering a concrete path for manufacturers to deploy these technologies without building custom solutions from scratch.

The partnership details were first reported by Interesting Engineering.