German English

Ensenso C67 for Real-Time 3D Parcel Dimensioning

Used products:C67

Introduction

Progressive Robotics develops AI-powered robotic palletizing solutions for warehouse and intralogistics operations. Their systems are designed to automate complex palletizing tasks, including the palletizing of mixed parcels without prior sequencing.

For this application, accurate 3D perception is essential. Incoming parcels vary in size, shape, surface material, and orientation, so the system needs reliable real-time dimensioning before deciding how each parcel should be handled and placed on the pallet.

To improve the performance of this box-dimensioning pipeline, the company recently upgraded to the Ensenso C67, supplied by Optonic.

Application Overview: Real-Time 3D Dimensioning for Mixed Palletizing

In on-the-fly mixed palletizing applications, parcels are measured as they arrive. The camera captures 3D data, which is then used by the software stack for box segmentation, surface estimation, and final dimension calculation.

This information feeds into the wider palletizing logic, helping the system (AnyStack Palletizer) make real-time decisions without requiring the parcels to be pre-sorted or sequenced.

In this type of application, the quality of the point cloud directly affects the stability of the entire process. Cleaner and more consistent 3D data leads to more reliable measurements, better repeatability, and fewer uncertainties for the downstream software.

Challenge with the Previous Setup

The previous camera setup was functional and could detect incoming parcels reliably. However, as the application moved closer to production requirements, its limitations became more visible.

The main challenge was not detection, but measurement consistency. Reflections, noise, and frame-to-frame variation could affect the quality of the acquired 3D data. This, in turn, could influence downstream stages such as box segmentation, surface estimation, and final dimension calculation.

For mixed palletizing, even small variations matter. The system needs to process different parcel types continuously and make reliable decisions in real time. This made point cloud quality and measurement repeatability key priorities for improvement.

Solution: Upgrading to the Ensenso C67

The Ensenso C67 was selected because it improves the areas that matter most for our production use case: point cloud quality, measurement repeatability, and deployment robustness.

The new camera provides cleaner and more stable point clouds, with better resistance to reflections and other noise sources commonly found in industrial environments. This gives the processing stack a stronger input signal and reduces the amount of uncertainty that needs to be handled later in software.

Integration was also an important factor. The camera comes with a dedicated mount, which simplifies installation, positioning, and mechanical alignment on the machine. The C67 also introduces hardware updates that support a more robust industrial setup, including an M12 IO connector for improved robustness, mechanical stability, and operational safety.

From a software perspective, the integration remained clean and modular. The camera data flows into the existing box-dimensioning stack without requiring a major redesign of the pipeline.

Validation Results:
Previous Camera vs. Ensenso C67

To validate the upgrade, Progressive Robotics compared the previous camera setup with the Ensenso C67 across 20 different boxes and 120 total measurements, with 60 samples per camera.

Both cameras successfully detected all samples, so the comparison focused on measurement quality rather than basic detection capability.

 

Metrik Previous Kamera Ensenso C67 Result
Samples tested 60 60 Equal test coverage
Detections 60 / 60 60 / 60 Both detected all samples
Mean error 2,1 mm 1,4 mm Improved with C67
Median error 2,1 mm 1,3 mm Improved with C67
Length error 2,4 mm 1,5 mm Improved with C67
Width error 2,2 mm 1,0 mm Strongest improvement
Height error 1,7 mm 1,7 mm Similar performance
Boxes with lower average error 6 / 20 14 / 20 C67 more consistent overall

 

The Ensenso C67 reduced the overall mean dimensioning error from approximately 2.1 mm to 1.4 mm, while the median error improved from approximately 2.1 mm to 1.3 mm.

The strongest improvement was observed in width measurement, where the average error was reduced from approximately 2.2 mm to 1.0 mm. Across the 20 tested boxes, the Ensenso C67 delivered lower average error in 14 cases.

Feedback from Progressive Robotics

For the team, the main benefit of the Ensenso C67 is that the box-dimensioning pipeline now starts from a cleaner and more reliable 3D input. This improves repeatability, supports more stable segmentation, and helps the system produce more consistent dimensioning results in real production conditions.

The upgrade also made deployment easier. The dedicated mounting and industrial hardware updates support a more robust setup, while the software integration remained straightforward and modular. The updated model variants also provide more flexibility for application design, including smaller stereo vergence angle options that reduce shadowing near large object edges and help capture object edges more completely.

“As we continue developing AI-powered robotic palletizing systems for warehouse automation, robust 3D perception will remain a key part of our technology stack. The Ensenso C67 supports this direction by giving our software and robotics systems higher-quality data from the first step of the process” said Marios Kiatos, CTO at Progressive Robotics.

Conclusion

The upgrade to the Ensenso C67 shows how better sensor quality can translate directly into better system-level performance.

For Progressive Robotics, cleaner point clouds, improved measurement repeatability, easier deployment, and a more predictable dimensioning pipeline all contribute to a more reliable mixed palletizing solution.

Miniar Brenner
Director Business Development & Sales Ensenso
This website is using cookies to provide a good browsing experience

These include essential cookies that are necessary for the operation of the site, as well as others that are used only for anonymous statistical purposes, for comfort settings or to display personalized content. You can decide for yourself which categories you want to allow. Please note that based on your settings, not all functions of the website may be available.

This website is using cookies to provide a good browsing experience

These include essential cookies that are necessary for the operation of the site, as well as others that are used only for anonymous statistical purposes, for comfort settings or to display personalized content. You can decide for yourself which categories you want to allow. Please note that based on your settings, not all functions of the website may be available.

Your cookie preferences have been saved.