Why 3D Inspections Are Finally As Easy As 2D Inspections

machine vision camera lens up close behind checkmarks and x's

Most manufactured products are three-dimensional (3D) objects, so it makes sense that automation systems designed to operate in all three dimensions are more effective at assembly, quality assurance and other industrial tasks. Today, 3D machine vision systems typically perform three basic tasks: guiding automation (e.g., robots) to pick up objects, inspecting objects for defects, and gauging objects as part of an assembly process. 

Flexible 3D laser scanning technology solves challenging applications

Today, 3D machine vision designers have multiple options at their disposal, including laser scanning, stereoscopic, and time-of-flight (ToF) solutions. Of these three solution sets, laser scanning is by far the most common, providing the fastest, most accurate, and most cost-effective 3D data acquisition solution. 

The three primary types of laser scanners are: the laser profiler, the displacement sensor, and area-scan systems. Laser profilers generate surface maps by stacking individual profiles into a continuous image. Displacement sensors can generate true 3D point clouds of the entire object for better accuracy and usability than basic height maps.

The individual strengths of each 3D laser scanning solution make them attractive for a wide range of applications in automotive (flush & gap, part inspection, robot guidance), consumer products (counting, packaging, verification), food and beverage (inspection, counting, sorting, packaging), and electronics (presence-absence, quality assurance).

Three common challenges to 3D

3D vision solutions face three main challenges: The need to measure in three dimensions, including rotation around each axis for six degrees of freedom; the lack of true 3D tools; and the limitations of embedded processing and their impact on computing-intensive 3D applications. 

Measuring objects with true six degrees of freedom challenges 3D vision systems because most 3D systems use height maps rather than true 3D point cloud scans. With height maps, height information is encoded in color, similar to the way temperatures are encoded in a thermal image. Unfortunately, humans don’t see height in color, making these solutions hard to understand and program. Most 3D applications call for some level of movement, requiring the 3D vision system to accommodate rotation, tilt, and tilt direction. And finally, measurement accuracy can suffer when using 3D surface maps versus true point clouds because only point clouds represent the object as it appears in the “real world.”

3D As Easy As 2D-2D height map and 3D map comparison

The lack of true 3D vision tools means many 3D solution developers spend time trying to solve 3D applications using 2D tools. This results in less accurate data, higher investments in engineering and longer development and run-time operations.

Finally, 3D vision solutions process a lot of data! For this reason, most 3D solutions depend on industrial PCs or expensive industrial PC processing units. These processors complicate cabling and retrofits, especially for robot-mounted solutions that already bring additional concerns about cable wear and the need to limit weight at the end of the robotic arm. 

Building a better 3D solution

In light of the growing need for 3D vision solutions, Cognex has focused on solving the primary challenges to widespread deployment. The new 3D-L4000 solution, for example, uses a speckle-free laser to generate a full 3D point cloud instead of height maps and provides detailed information for each rotation and tilt. 

New 3D solutions also present the full 3D-rendered point cloud to the operator, allowing them to rotate, tilt, and spin the object in any direction. This makes it easier to correlate features on the part with features on the display.

Cognex developed vision tools that work directly on the 3D point cloud, providing higher accuracy, and expanding the types of inspections that can be performed. Better yet, because inspections are in 3D, users can immediately experience how the vision tools operate on the actual part or component. The 3D-L4000 offers multiple dedicated 3D algorithms, including Blob3D, PatMax 3D, Edge3D, ExtractSphere 3D, ExtractCylinder 3D, and more. All these tools are accessible through In-Sight’s patented spreadsheet-based user interface for intuitive programming and operation.

2D and 3D point cloud object scan

Today, using advanced 3D inspection systems, customers are able to deploy simple, non-contact solutions to eliminate defects and recalls. For instance, auto manufacturers can use 3D vision systems for brake pad inspection, verifying the gap width, the angle of the beveled edges, and the rivets to make sure that the friction material is secured correctly. These systems also perform 2D inspection, like checking reading date, lot codes, and labels that have been applied. This is only one of many applications that compact, easy-to-use 3D solutions are making possible. 

Mary Romano

Manager, Global Content Marketing, Cognex

A self-proclaimed marketing maverick, Mary has experience in a variety of industries and marketing activities. Her career began in the world of defense and aerospace, then shifted to collaborative robotics and now machine vision. She drives brand awareness, nurtures customer relationships, and elevates the impact of sales through compelling content. Outside of the office, Mary enjoys baking (and eating) desserts, running outdoors, and exploring new places. She is also a self-published author of a children’s book about her dog, Doma.

More Posts on


Join MyCognex


Cognex representatives are available worldwide to support your vision and industrial barcode reading needs.

Contact Us