“After 35 years of AI research, one of the important lessons people have learned is that hard problems are simple, and simple problems are hard.”

— Steven Pinker

Cognitive scientist

In the industrial world, there is a “simple” problem that is really troubling – catching things.

In the West home, there is such a group of people who are committed to this “small matter”.

Artificial intelligence can bring more intelligent control systems to industrial robots.

At present, online shopping has become the norm, all kinds of e-commerce has flourished, and the number of packages under the shopping carnival is amazing. According to the monitoring data of the State Post Bureau, as of June 24, 2023, the national express delivery volume this year has reached 60 billion pieces, 172 days ahead of the 60 billion pieces in 2019, and 34 days ahead of 2022.

A wide variety of different shapes of items gathered together, behind the scenes are thousands of staff sorting, packaging, handling and other work, the warehouse of the busy scene can be imagined.

Can we use the power of technology to ease some of the burden on people? The answer is yes. For example, the AGV car, called the “warehouse handling efficiency artifact”, can replace manual rapid handling of goods and is widely used in the field of intelligent warehousing.

The robot handling system of Shanghai Siemens Switch Co., LTD. (SSLS) is an important part of intelligent logistics and warehouse management in the factory.

While handling has become increasingly automated, sorting is still a labor-intensive operation that requires a large number of employees and takes an extremely long time.

You may ask, why can’t sorting be automated when it is simply a matter of picking up the corresponding items and putting them back down? The reason lies in the complexity of the human brain. For the “small matter” of grasping something, the human brain will quickly identify the grasping point of different objects, and the robot does not have such cognitive ability.

If the robot arm is the “arm” of an industrial robot, it enables it to grasp the action; Then the camera is the “eye”, so that the robot can recognize and track the object based on visual technology, and achieve “hand-eye coordination”. However, the robot lacks the “brain” of independent thinking, and cannot adjust the grasping strategy independently and carry out disorderly sorting in a mixed and obscured environment. Even by creating established programs, industrial robots can only recognize the grasp point of “known” items, that is, the placement of fixed, shaped objects.

In fact, not only in the logistics industry, but also in the warehouses and production lines of all walks of life, disorderly sorting tasks can be seen everywhere, and more than 90% need to rely on manual completion, and the cost can not be naturally reduced.

Siemens AI solutions give robots a “brain” that can think for themselves.

In response, colleagues at Siemens’ factory automation division proposed an innovative technological route: let AI learn the basic geometric features of a large number of objects. This is different from the traditional point-to-point teaching grasping scheme and the grasping scheme based on product appearance template matching, which can enable the system to fundamentally learn the skill of “grasping” and achieve “zero training” plug and play. Siemens’ deep learning vision software SIMATIC Robot Pick AI was born. It can be paired with any robot arm and industrial camera, so that industrial robots can easily “pinch” stacked chaotic and diverse structures of objects, and truly achieve disorder grasp!

 

SIMATIC Robot Pick AI is characterized by openness, modularity and simplicity. The user does not need to perform model matching, no manual training data, and is ready to use, and the intelligent system can recognize the grab point in real time in the running state. Not only that, the system is perfectly compatible with the SIMATIC family and TIA Boto software, and it takes only ten minutes to install!

 

SIMATIC Robot Pick AI, quickly achieve flexible, accurate and efficient random picking and sorting.

The “small matter” of robot grasping is of great significance to the future of industry, not only to mention efficiency and accuracy, but also to liberate a large number of workers from high-intensity repetitive labor. With the increasing aging and labor shortage, this “small matter” will also release greater potential to make science and technology serve humanity and make human work more valuable!