‘Adaptive Manufacturing’ – New Smart Manufacturing Era
As the world rapidly embraces Generative AI and Agentic AI, a new frontier is emerging – Physical AI, where artificial intelligence meets the physical world through robotics and automation. At the forefront of this evolution is 3D Infotech, a pioneer in adaptive manufacturing and quality control automation.
Following the company’s participation at the 2025 NVIDIA GTC AI Conference, Metrology News sat down with Rohit Khanna, Founder and Chairman of 3D Infotech, to explore how their innovations in Physical AI are reshaping smart manufacturing and setting the stage for a more intelligent, responsive production future.
Q: You’ve coined the term “Adaptive Manufacturing” Can you tell us what that means in practice, and how it goes beyond traditional factory automation?
A: Traditional factory automation is centered around high-volume production for a limited set of parts. The typical process of setting up parts for welding, deburring or any other robotic applications is to program a robot with a specific path and execute it on parts as they are presented to the robot in a specific coordinate frame. The robot programming is either done offline using a simulation software such as DELMIA or RoboDK, or manually by jogging the robot to specific positions. This can be tedious and time consuming. It often requires a high skilled worker that has been trained in robotics and engineering in order to create work instructions for the robot. This methodology creates a set-it and forget it mentality and removes any flexibility from the process to account for any part variation or change in the parts that are being run through the robotic cell.
Adaptive Manufacturing is a methodology where metrology is embedding in the robotic cell and using Physical AI, the system can adapt to changes that might occur in the production plan. You may present a different part number to the cell, in a different location (within reason), or even have the same part that has variation from the golden part that a robotic cell might have been trained for.
Q: How does Adaptive Manufacturing redefine how manufacturers approach flexibility, scalability, and autonomy in production environments?
A: Conventional robotic applications are highly structured and quite inflexible. Compounded by the overhead of programming parts using highly skilled workers, has resulted in a resistance of adopting robots in applications where there is a high mix of parts and low volume of production. Adaptive Manufacturing offers CNC operators and other shop floor workers alike to place any part, anywhere in the work cell and the robot will dynamically locate the object, classify it and generate a new robot path for the operation to be performed. This high level of flexibility greatly reduces the burden to prepare parts for the robotic cell.
Q: In what ways is Adaptive Manufacturing the “missing link” in fully autonomous factory cells?
A: Imagine a factory that produces hundreds of part numbers on a weekly or monthly basis but with a low enough volume of production that does not justify having a dedicated robot programmer to create, test, optimize, deploy and maintain all these parts for a robotic cell. Adaptive Manufacturing allows regular operators to pull our any part that is ready for inspection, deburring, sanding, polishing or any other “light manufacturing” task and present it to the AI-driven robotic cell. The robotic cell is intelligent enough to realize that a part has been placed for the task to be performed and simple does it, without the need for the operator to press any buttons to initiate the process. Of course, the presenter of the part can be another robot dedicated to loading and unloading parts, further removing the need to apply human labor. This is the bridge or missing link between human-driven machine instructions that have prevented fully autonomous robotic systems, and the factory of the future where autonomous robotic cells work on tasks by themselves.
Q: How does Intelligent Metrology enable or accelerate the development of fully adaptive manufacturing cells?
A: Metrology is the science of measuring objects to a certain precision. By addition AI to the mix, this gives birth to Intelligent Metrology. 3D Infotech’s first AI-driven robotic system is exactly that. Operators are able to inspect parts automatically and visually see defects on the parts using an automatic projection of areas outside of tolerances with respect to CAD models. PolyWorks software is used to do the metrology comparison, and the operator does not even need to press a single button. Removing the need for human involvement gives manufacturing companies the opportunity to accelerate their development and deployment of fully adaptive manufacturing cells.
Q: Your technology integrates NVIDIA’s AI Accelerator and Universal Robots—how do these elements work together to enable intelligent decision-making in real-world manufacturing cells?
A: NVIDIA provides highly powered computational elements that allow us to store thousands of Machine Learning training models that are synthetically generated using 3D Infotech’s proprietary software, Streamline X. This software utilizes Universal Robot’s PolyScope X platform which provides a Realtime link between the intelligence of the AI-driven software and the physical arm of the collaborative robot. We have not been able to accomplish this task with any other robot brand primarily because the foundational hardware and software has been missing.
Q: What makes 3D Infotech’s approach to AI-driven automation different from other robotics or automation companies?
A: The perfect match between Universal Robots’ platform and 3D Infotech’s software expertise has been the key differentiator. Our experience and knowhow in metrology and robotics a variety of manufacturing processes including castings, forgings, sheet metal stamping, plastic injection molding and more has enabled us to build a workflow that can give rise to tens of adaptive manufacturing vertical solutions.
Q: You previously mentioned that the technology is already expanding into other processes like sanding, deburring, welding, and assembly. What has been the biggest technical hurdle in applying the same intelligence across such diverse applications?
A: Having access to ML models that have “tribal knowledge” about these specific processes is still missing. With the advent and rising adoption of agentic AI, 3D Infotech hopes to tap into these knowledge bases real-time and apply them across these diverse application sets.
Q: How modular is the Adaptive Manufacturing platform—can customers start small and expand over time, or is it a comprehensive deployment from day one?
A: In its current state, the Adaptive Manufacturing platform is not ready for mass deployment. With beta customers engaged currently, and further maturity to the platform we hope to start commercially deploying our adaptive manufacturing solution in 2026. Our intention from day 1 has been to have a modular and scalable architecture, one where customers can start with one cell and expand it to larger part sizes or expand the number of cells easily.
Q: The concept of “Physical AI” is gaining momentum, especially after your spotlight at NVIDIA GTC. How do you define Physical AI, and how is it distinct from Generative or Agentic AI?
A: Physical AI is the next big AI trend. It is gaining momentum with each passing day. The emphasis for Physical AI has been in humanoid robots, but 3D Infotech is focused on a subset of this using robotic arms, where the return on investment is clearly defined, and attainable in a reasonable time frame. Generative and Agentic AI are supporting technologies that will help Physical AI advance into areas that are seemingly impossible to attend to. At this point, we can think of them as building blocks that have led to the Physical or what we call Robotic AI revolution.
Q: If you were to look five years ahead, what would a fully Adaptive Manufacturing cell look like—and what impact would it have on productivity and the workforce?
A: 3D Infotech is a global company, and our 2030 vision statement focused on the supporting our deployments in key regions such as United States, Mexico, Europe and Asia Pacific. As we expand our technology portfolio, we will bring the benefits of Adaptive Manufacturing to companies worldwide, delivering increased productivity by up to 5x while reducing dependency on workforce by at least 2x. In the past, many companies have hesitated to adopt robotic automation due to increased overheads, so we hope to make it easier for them to take advantage of the intelligence and productivity gains for a variety of production processes.
For more information: www.3dinfotech.com
