Within the industrial value chain, there is a multitude of applications in which robots automatically grasp specific objects. In classic pick-and-place applications, objects are picked up at one location (pick) and placed at another (place).

Automated industrial workflows can no longer do without robot-assisted gripping processes. While pick-and-place applications require objects to be securely gripped and placed elsewhere, bin picking is more complex: the objects to be gripped are usually lying in disorder and sometimes on top of each other in containers, which makes automated gripping difficult or nearly impossible. Here, the new Deep 3D Matching technology from MVTec comes in. It recognizes the exact position of objects even under difficult conditions, making picking from a box safe, fast, and efficient.

Within the industrial value chain, there is a multitude of applications in which robots automatically grasp specific objects. In classic pick-and-place applications, objects are picked up at one location (pick) and placed at another (place). This process is relevant in all stages of production, such as manufacturing, assembly, quality inspection, and packaging. For example, robots pick components from a conveyor belt or pallet, separate them, and prepare them for further production steps. Pick-and-place applications are also used in intralogistics, such as when picking items from a warehouse.

Other typical applications include placing components on circuit boards in electronics manufacturing or assembling components into a final product on assembly lines. At the heart of many pick-and-place systems is machine vision software combined with one or more cameras for reliable detection of the objects to be gripped. Pick-and-place technologies are fast and precise and therefore increase production capacity and relieve employees of monotonous and strenuous manual handling tasks.

Bin picking - the supreme discipline of automated gripping
Bin picking is a special type of automated gripping. It is considered the supreme discipline within pick-and-place processes and places high demands on the entire process: the objects to be gripped are usually lying in the bin in a disorderly manner and sometimes overlap. Furthermore, the containers often contain different objects of varying sizes and shapes. The robot must pick up all these objects safely and with high precision. An essential prerequisite for this is that it can reliably recognize the corresponding parts and locate them with pinpoint accuracy by utilizing powerful machine vision software.

MVTec Software GmbH has developed various machine vision methods for realizing both regular pick-and-place and bin picking applications: a commonly used method is shape-based matching. This method enables particularly accurate and robust localization of a wide variety of objects in real time. The decisive advantage of this technology is that it delivers very robust and sub-pixel-accurate recognition rates even under difficult conditions - for example, with rotated, scaled, or partially covered objects. These special properties make shape-based matching ideal for demanding bin picking applications.

3D matching ensures robust recognition rates
Other important technologies include 3D-based vision methods such as perspective matching. This allows the position of arbitrarily shaped, flat, or tilted objects in three-dimensional space to be determined with high precision. Surface-based 3D matching and MVTec's 3D Gripping Point Detection technology deliver similarly good results. The latter can also be used to reliably detect and grasp objects whose appearance is not known in advance. The method can identify possible gripping surfaces for vacuum suction cups on the objects so that the robot can pick them up precisely.

The most reliable results in bin picking applications are achieved using methods based on artificial intelligence. MVTec has developed a new method called Deep 3D Matching. This feature combines deep learning algorithms with rule-based methods, enabling unprecedented and highly robust recognition rates. A special feature here is that just a few 2D images are sufficient for accurately locating objects in three-dimensional space. This saves costs, as only inexpensive 2D cameras are needed here. The more cameras are used, the more robust the recognition rates become. Yet another advantage is that the required number of cameras can be added flexibly. Only a few minutes of calibration work are required, which significantly reduces the effort involved. One more benefit is that only synthetically generated image data can be used for training as well. These images can automatically be produced in large quantities and at low cost using the CAD data of the objects to be recognized. Furthermore, users save time and money, as the images do not need to be labeled. In addition, only a small amount of parameterization is required to achieve robust recognition rates.

Proven matching technologies for reliable gripping processes
"Industrial bin picking and pick-and-place applications place high demands on the technologies used in this context. To reliably grasp and pick up objects, the robot must be able to precisely recognize and locate them in three-dimensional space. With our diverse and proven matching technologies, we are well positioned here, and Deep 3D Matching in particular, offers users significant added value, such as high speed and outstanding recognition rates. Moreover, companies benefit from significant cost advantages thanks to synthetically generated data and minimal parameterization effort," concludes Thomas Huenerfauth, Portfolio Manager Vision Technologies at MVTec.

MVTec will be presenting numerous machine vision technologies for automating pick-and-place and bin picking applications at this year's automatica in Munich from June 24 to 27, 2025, at booth B5.303.


About MVTec Software GmbH
MVTec is a leading manufacturer of standard software for machine vision. MVTec products are used in a wide range of industries, such as semiconductor and electronics manufacturing, battery production, agriculture and food, as well as logistics. They enable applications like surface inspection, optical quality control, robot guidance, identification, measurement, classification, and more. By providing modern technologies such as 3D vision, deep learning, and embedded vision, software by MVTec also enables new automation solutions for the Industrial Internet of Things aka Industry 4.0. With locations in Germany, the USA, France, Benelux, China, Taiwan, and South Korea as well as an established network of international distributors, MVTec is represented in more than 35 countries worldwide. www.mvtec.com