Robots have a long-standing blind spot, and it is not a software bug. They simply cannot see glass, polished metal or jet-black surfaces. A team at Fraunhofer Institute for Applied Optics and Precision Engineering IOF says it has solved that problem with a new measurement system that combines infrared laser projection with thermography, and has now put the sensor to work on a robot for the first time.
The result matters because anyone who has tried to automate inspection, pick-and-place or assembly lines involving transparent components knows the frustration: the moment a glass part enters the cell, the vision stack breaks down and projects stall.
Key facts
- Fraunhofer IOF researchers built a sensor that detects transparent, reflective and light-absorbing surfaces in 3D.
- The system, called Glass360Degree, uses infrared heating plus two thermal cameras instead of paint or stickers.
- It has now been tested on a robot in an industrial setting for the first time.
- Target sectors include semiconductor, automotive and aviation manufacturing.
- The team presented it at Hannover Messe and Control trade fairs in 2022.
Background: why transparent objects break automation
Glass, glossy metal and dark polymers are called "uncooperative surfaces" in the vision world. Standard 3D sensors rely on visible light, which either passes through glass or bounces off shiny metal in unpredictable directions. The usual workaround—spraying a temporary matte coating—slows cycle time and adds a consumable. For high-mix, low-volume lines, that workaround is often uneconomic, which is why many automation projects stall at the planning stage.
Fraunhofer's approach sidesteps the problem by heating the surface locally with a CO₂ laser and watching how it cools. Two thermal cameras capture the temperature pattern from different angles, and specialized software reconstructs the 3D geometry from the heat signature. No markings required. The research group, led by Dr. Stefan Heist, has been working on 3D measurement methods for years and partnered with Ilmenau University of Technology to translate the measurement data into robot control signals.
Who should care
| Sector | Current limitation | What Glass360Degree changes |
|---|---|---|
| Semiconductor manufacturing | Transparent wafers and glass components resist standard sensing | Enables in-line metrology for parts previously requiring manual handling |
| Automotive | Dark polymers and reflective trim on fixtures block robot guidance | Opens previously stalled automation projects for glass and polished metal assemblies |
| Aviation | Composite and coated surfaces create mixed-surface challenges | Allows consistent 3D capture across varied surface types without prep work |
| Quality assurance | Matte spray application adds cycle time and consumables | Eliminates marking step, reducing per-part cost for high-value items |
Our analysis
The headline is not "a new camera." It is that a class of previously uneconomic automation is now technically viable. Cells that used to need a human for visual checks on glass or polished parts become candidates for unattended operation. That shifts the bottleneck from sensing to integration: the downstream system must receive reliable 3D coordinates fast enough to maintain line takt.
The trade-off is thermal: the system introduces a small heat pulse (typically less than 3 °C above ambient) into the measured surface. For most industrial parts this is harmless, but teams will need to verify thermal limits for sensitive components. The temperature difference is brief and localized, but cycle-time impact depends on the object's thermal mass and the required measurement resolution.
Watch for published cycle-time benchmarks from the trade fair demonstrations. That number determines whether this stays in the research phase or moves into procurement conversations. The balance between resolution detail and measurement duration was identified as a key development area, and industrial adoption will depend on how Fraunhofer optimized that trade-off.
What to watch next
Hannover Messe was the first public test. If published benchmarks show cycle times competitive with conventional 3D sensors, expect pilot inquiries to increase. Until then, treat the demonstrations as evidence of technical viability, not as a signal to begin integration planning. The sensing capability is real; the integration pathway still needs benchmarking data that was not available at the time of the initial demos.
Key takeaways
- Transparent-surface sensing is no longer a research-only topic; it has operated on a real robot.
- The main advantage for industrial teams is unlocking automation projects that stalled on "vision infeasible."
- The system requires no temporary markings, which removes a consumable cost and a process step.
- Watch for cycle-time benchmarks before committing to integration timelines.
- Thermal limits on sensitive components should be verified during any pilot evaluation.
Frequently asked questions
What is the new sensor called?
The Fraunhofer IOF team refers to it as the MWIR 3D sensor, also branded Glass360Degree.
Which industries is it aimed at?
Semiconductor, automotive and aviation manufacturing are the named target sectors.
How does it see glass?
It heats the surface locally with a CO₂ laser and uses two thermal cameras to read the temperature pattern, then reconstructs 3D geometry from the heat signature.
Does it need paint or stickers on the part?
No. The system works without temporary markings, which is one of the main advantages over conventional 3D sensors.
What was the thermal impact on measured objects?
The temperature difference between heated and non-heated surfaces is typically less than 3 °C, and the thermal energy introduced is low enough that the object is not damaged.