Kiwi-Eye, the first system tested for safety and efficiency in the workplace

Kiwi-Eye, the first system tested for safety and efficiency in the workplace


It is the first selective detection system with artificial vision for the industry: Kiwi-Eye, the creation signed by Kiwitron from design to assembly, is about to obtain the patent that protects its uniqueness. 

Kiwi-Eye is a computer vision system applicable to all mobile and industrial machines, capable of detecting distances and depths, of recognizing and distinguishing human beings, other machines, objects and signs. But the list does not end here: it can be extended without limits, depending on the needs. 

How? We tell you about the technology behind the flagship product of the Bolognese company.

The patent application

Kiwi-Eye is the first distance detection system with artificial vision tested, applied specifically to the needs of mobile and industrial machines and at the service of safety and efficiency. For this reason, in the summer of 2020, Kiwitron immediately moved to protect its product, filing a patent application. Subsequently, in order to extend the protection on this unique and innovative product, Kiwitron also filed a second application.

This describes in detail all the characteristics of the Kiwi-Eye product, but aims at a much wider protection, aimed at a device consisting of an RGB camera, a depth camera and a control unit useful for establishing the nature and distance of the objects detected thanks to to artificial intelligence algorithms. Recently, the application extended to two patents.

In a short time, the company will obtain exclusivity on the solution that promises to transform mobility in the workplace and the concept of a semi-automatic industrial machine. All with the mission of increasing business productivity by decreasing security risks. 

The unique benefits of Kiwi-Eye

Kiwi-Eye is the state-of-the-art solution for the selective recognition of obstacles and for the mitigation of the risk of accidents. The detection reaches a distance of 25 meters and boasts an accuracy to the centimeter. The immediate benefits of this hi-tech products are: 

  • thanks to the developed neural networks in house, Kiwi-Eye can learn to recognize and report any type of object;
  • it is designed and produced entirely in the Sasso Marconi headquarters, where customer assistance and the commercial area are always ready to listen to the customer's needs and act as a link with the Kiwitron laboratory;
  • thanks to the integrated depth chamber, the distances indicated are exact and not the result of an approximate estimate;
  • Kiwi-Eye can identify high-risk areas in the workplace thanks to the automatic collection of data on near-miss (near misses). The Kiwitron software for fleet management can be viewed in the Heatmaps area of Kiwisat.

Thanks to these features, Kiwi-Eye eliminates the need for anti-collision tags in the warehouse and on the construction site. It is possible to configure 3 alarm zones based on customizable distance thresholds and based on the vehicle's running signal. Furthermore, it is able to communicate / interact with the machine (for example, it can activate its slowdown) and does not require a CPU HUB.

But Kiwitron's competitive advantages lie not only in the great technological content of its product, but also in the structure and team.

In fact, the company has its only headquarters in Sasso Marconi, in the Bologna area. This is where all the solutions for safety and efficiency come to life, where the customer can always have direct contact with the company. 

Do you want to know more?

KiwiEye sistema di rilevamento con intelligenza artificiale

Kiwi-Eye: how industrial computer vision works

Kiwi-Eye is able to constantly improve its performance by transmitting data to Kiwitron servers. The amount of encrypted data is processed through algorithms of deep learning, the most advanced artificial intelligence technique.

In fact, Kiwitron guarantees that the models of computer vision are able to recognize certain obstacles, in potentially always different circumstances. When a Kiwi-Eye mounted on the front of a truck is in a warehouse full of operators and vehicles, its job is to identify only a few risks and not trigger alarms or slowdowns unnecessarily. But what technology does Kiwi-Eye use?

Before leaving Kiwitron productions, the Kiwi-Eye goes through a phase of training, learning at the algorithm level. To ensure that the system learns to recognize objects useful to customer needs, Kiwitron's R&D teams apply automatic procedures for collecting, pre-processing and storing data in real time. On this data it is possible to implement deep system learning.

The technique of deep learning of the Kiwi-Eye system is called, in technical jargon, computer vision - or "artificial vision". Machine vision is the area of AI dedicated to processing and understanding images and videos.

However, for Kiwi-Eye it is not just a matter of recognizing the pattern of an image, but also of extracting specific information for specific industrial purposes. For this, the Kiwi-Eye AI recognition tasks are performed through complex neural networks, the convolutional neural networks (CNN).

CNNs are a specific type of neural networks: they are “trained” on very large datasets and are able to apply problem solving techniques. 

The use of CNN radically improves the standard Computer Vision algorithms, allowing the identification of patterns between images in total autonomy.

CNNs eliminate the need to define image characteristics with human intervention. Kiwi-Eye “learns” to decipher what it sees automatically, producing highly accurate recognition results.

Leave a Reply

Your email address will not be published. Required fields are marked *