Breaking the high-mix bottleneck with intelligent automation

TRUMPF SortMaster Vision (Fanuc R2000ic-210L) © TRUMPF

For more than 3 years, combining deep industry expertise and passion for intelligent automation, TRUMPF and Intrinsic have reached substantial milestones together with the goal to build a new generation of intelligently adaptive robotic solutions. Together, they share a vision that AI-enabled robotics will unlock new levels of productivity for machine users, in affordable, seamless ways that make sense on the shop floor. Today, fabricators in sheet metal face relentless pressure to optimize their processes for greater efficiency and adapt to evolving customer demands, all while maintaining their existing systems, quality, and output. The need for high-mix, low-volume automation has never been higher or more valuable for growth.

This collaboration has redefined the sorting workflow for TRUMPF SortMaster, a high-speed automated system designed to remove, sort, and palletize laser-cut sheet metal parts. It has automated a largely non-automated step of sorting parts after a 2D laser cutting machine. Most of the time, especially in high-mix, high volume production, this tiring task is performed manually. The sorting process is split into two phases. The TRUMPF SortMaster Station first separates the cut metal parts from the remaining skeleton.

This clears the way for the TRUMPF SortMaster Vision, which utilizes a Fanuc R2000ic-210L to intelligently sort and organize these freed parts onto pallets and bins. Developed and deployed with the Intrinsic platform, TRUMPF’s SortMaster Vision transforms rigid, manually programmed systems into an intelligent 'see-and-sort' automated process. The Intrinsic Vision Model (IVM) - an award-winning AI perception model - detects multiple objects and enables precise robot movement to pick every part successfully. The system leverages one shot, CAD-based deep learning to identify and stack parts autonomously, enabling real-time adaptation with zero manual programming -increasing system productivity and reducing cycle times.

The challenge: Many TRUMPF customers operate in high-pressure machine shop environments, requiring the agility to produce small, custom batches on demand. The flexibility to recognize a wide variety of part geometries without any training times, manual programming or intervention is a huge opportunity. TRUMPF looked for a specialized machine vision foundation model that could seamlessly integrate into the core of their solution.This level of pose and part detection precision is essential when parts overlap. Sometimes the sheet metal parts are overlapping by just a few millimeters, and the vision model has to decide exactly which part is on top so that it doesn't pick up a second part accidentally.

The Solution

With the addition of the Intrinsic Vision Model, TRUMPF’s SortMaster Vision stands out because it leverages AI to deliver:

• AI part recognition for zero-programming automation: The system was trained on an extremely wide range of sheet metal parts. Therefore, it does not require any customer-specific training. The CAM software simply provides the CAD models, and the robot knows how to pick and sort the parts using highly efficient motion paths. This removes the need to manually program or reprogram the robot, when new parts are added.
• Maximum process reliability through high precision: The Intrinsic Vision Model (IVM) excels at "segmentation" and "pose estimation." It identifies individual parts even when they overlap by just a few millimeters, preventing accidental double-picks or collisions in complex scenarios.
• Process optimization with unrivaled flexibility: The system recognizes a vast range of part geometries. This makes it perfect for dynamic, high-mix production and fulfilling urgent machine shop orders.
• Unattended production for solid reliability and less downtime: By separating parts before sorting, the system avoids common issues like parts tipping or jamming during skeleton removal. This results in significantly higher success rates for "lights-out" operations.

This allows robots to autonomously identify and sort sheet metal parts using CAD-based deep learning, in a low-cost way, eliminating the need for manual programming.

The benefits of building with Intrinsic

The Intrinsic Vision Model (IVM) is an award winning, industrial-grade foundation model designed to automate complex perception tasks such as pose estimation, part/object detection, and tracking. Unlike traditional systems that require extensive manual programming or machine learning training/cost for every new object, IVM reasons directly with CAD models. This "CAD-native" approach enables zero-shot, sub-millimeter accuracy for new parts immediately upon deployment.

By utilizing specialized transformers and multi-view reasoning, the model achieves high precision using standard and affordable RGB cameras. Ultimately, IVM provides a versatile, "plug-and-play" solution that allows robots to handle high-mix production and tight-fit insertions with unprecedented reliability across diverse materials and lighting conditions.

Building with Intrinsic streamlines the entire robotics lifecycle. It combines Intrinsic Flowstate - a powerful web-based digital twin and developer environment - with ready-to-use AI capabilities, like the Intrinsic Vision Model. This allows system integrators to seamlessly develop, deploy, and manage adaptive solutions that run with industrial-grade reliability in daily operations.

Deployment and future outlook: Practical, Scalable AI Robotics

The success of the SortMaster Vision is being validated with real customers across industries. The system has been deployed by sheet metal manufacturers across Germany. In addition to reducing programming and offering greater flexibility, TRUMPF is convinced that this solution is more reliable than their conventional machine sorting process.

Positioned as a leading innovation partner, TRUMPF played a critical role in the long-term validation and testing of the IVM. This collaboration highlights TRUMPF’s prescience as an industry innovator, actively shaping the AI-driven tools that are now redefining benchmarks for productivity and flexibility, and cementing their status as a pioneer in AI-enabled manufacturing.

Tom Schneider, CTO of the Machine Tool division at TRUMPF, aptly summarizes the impact: "Customers can now automatically sort parts that couldn't be removed using conventional solutions. This innovative process is a game-changer, offering a fully automated solution for separating and sorting parts. Intrinsic's AI platform perfectly complements our vision for the future of manufacturing, where intelligent automation empowers businesses to achieve new levels of productivity and flexibility."

Intrinsic has a mission to democratize access to robotics for millions more businesses, manufacturers, and developers. Their partnership with TRUMPF and the successful deployments to their customers are exciting proof points of the potential of AI in industry. Both companies share a vision to make AI usable and valuable for machine tool users who want more productivity in practical ways. This collaboration paves the way for a future where exponentially more businesses can grow and scale with automation and AI.