Flexible, efficient, and above all reliable and safe: That’s what deployable person recognition in industrial production should look like. In the PROTECT project, Fraunhofer IKS and Fraunhofer IGCV have taken a big leap toward delivering these capabilities.
January 30, 2026
© iStock/Nano Stockk
PROTECT, the joint project by the Fraunhofer Institute for Cognitive Systems IKS and the Fraunhofer Institute for Casting, Composite and Processing Technology IGCV, is prototyping a camera-based, AI-powered person-detection system that surpasses traditional safety measures like light barriers and boosts overall system efficiency. The new safety layer reliably detects people and distinguishes them from other moving objects such as automated guided vehicles, enabling dynamic speed adjustments and faster cobot operation in a production cell—laying the groundwork for flexible, high-performance, and cost-effective manufacturing.
We are seeking feedback from industry and certification authorities, as the safety of AI components remains an open issue and is of great relevance to all stakeholders. At the end of 2025, researchers from the participating Fraunhofer Institutes presented the PROTECT project in an interactive workshop with users from mechanical and plant engineering, robotics, sensor technology, and regulatory institutions (TÜV) at the Fraunhofer IGCV technical center in Augsburg.
Introduced by Yaskawa's robotics specialist, an open discussion developed on equal footing, focusing on challenges and possible use cases.
Key questions included: How can we increase productivity, flexibility, and efficiency without compromising safety? What can be implemented in the near and medium term, how, and who do we need to do it?
We look forward to continuing the discussions and further developing our solution. Do you have any questions or suggestions? Would you like to talk to us without obligation? Then please contact Anna Guderitz, anna.guderitz@iks.fraunhofer.de, phone +49 160 99642798.
Currently, factory robots have relied on bulky, static safety measures—fences, light barriers, and slowdowns—that drive up costs, curb flexibility, and sap performance. Now, companies are working to remove restrictions and unlock more robot capabilities, all while maintaining worker safety in human-robot collaboration.
One of the primary safety concerns with AI systems is the possibility of false detections, which may lead to accidents erode trust, and negatively impact performance. . The certification maze—especially for AI components—adds to the challenge.
To address concerns – both on the safety and the performance side - , PROTECT deploys a hybrid approach that blends AI-based and rule-based methods to reliably recognize people and objects.
PROTECT demonstrates a lean use case: a camera-based system reliably detects people—even in tough conditions or untrained appearences—and differentiates them from AGVs. This enables zone-based, situation-aware speed control, slowing the robot only when a person comes near. With no collision risk, such as around an AGV, the robot can keep running at full speed.
The focus is on three key aspects that increase safety compared with existing solutions:
Conventional, non-AI motion detection adds an extra safety layer, crucial for spotting concealed individuals or those with unknown appearances, including people in protective gear.
See the next iteration of the PROTECT project with dynamic safety zones in action at the Transform trade fair on March 18–19, 2026, in Berlin. The Fraunhofer IKS team looks forward to welcoming you at the Fraunhofer ICT Group booth. For more information about Transform, visit:
Weitere Infos zur Transform finden Sie hier:
https://transform.show
Would you like to make an appointment? Please contact Anna Guderitz, anna.guderitz@iks.fraunhofer.de, phone +49 160 99642798.
The system automatically tunes the robot arm speed: Zone 1 runs at full speed, zone 2 slows when a person is detected, and zone 3 stops the arm when a person is present. When AGVs are present, the speed remains constant.
This innovative AI-based person-recognition system ramps up automation, strengthens safety for workers and equipment, and drives efficiency across modern production environments. Its greater flexibility underpins modular production that can rapidly adapt to shifting requirements and small-batch runs.
