when French technology is at the service of industry

WHAT PRESIDED IN THE CREATION OF ANOTHERBRAIN?

AnotherBrain was founded by Bruno Maisonnier, to whom we owe the famous humanoid robots Nao, Pepper and Romeo. After selling his company Aldebaran Robotics to the Japanese group SoftBank in 2015, he wanted to take a decisive step in the development of real intelligence that would endow machines with “emotions” and allow them to fully interact with the surrounding world.

Created in 2017, AnotherBrain is a French company that evolves in disruptive technologies. We are working on the third generation of Artificial Intelligence, known as general AI. Reproducing the functioning of the cerebral cortex, the “Organic AITMas we call it, learns without supervision and does not require “big data”.

Based in Paris and employing around sixty people, mainly researchers, developers and engineers, AnotherBrain wanted to master the software and hardware aspects of its solutions from the start. This dual expertise makes it possible to design the smallest possible chips that consume little bandwidth and energy.

Also based on this principle of frugality, our model works with very little input data. According to an edge computing approach, data processing is done locally and not from a centralized cloud. This reinforces the security of sensitive data. To develop, AnotherBrain raised 10 and 19 million euros in 2017 and 2019. The company was labeled FT120 in 2020.

WHAT MAKES ANOTHERBRAIN DIFFERENT IN THE COMPUTER VISION QUALITY CONTROL MARKET?

AnotherBrain’s first application was marketed in June 2021. PHOSPHOR® Quality is a turnkey vision, hardware and software collaborative quality control solution for any type of industrial process. From a 100% graphical and “user-friendly” interface, an operator independently creates an AI-based detection model from shots of “golden samples” of the object to be inspected online.

The device is equipped with a camera lamp, mounted on a robotic arm from the market leader to take these shots in an optimal way. Equipped with 300 LEDs, which can be switched on independently, the PHOSPHOR® Cam can rotate around the room to capture from the most relevant angles with suitable lighting to reduce reflection problems. The trajectory and illumination conditions recorded during the training phase are then reproduced for online quality control.

Unlike traditional Computer Vision systems that require large datasets for training, only about 30 images of conforming parts are needed to begin defect detection. This frugality in sampling makes it possible to launch a production line without having first constituted a complete photo library. Learning on correct parts frees the manufacturer from an exhaustive list of faults to be presented to the learning automaton and enables the system to detect “new faults” never before seen in production.

Once the model has been created, any part will be compared by PHOSPHOR® Quality to the ideal representation of a correct part. PHOSPHOR® Quality performs all types of integrity checks (presence or absence of components), appearance (dirt, scratches, impurities, discoloration, pitting) or structural checks such as deformations or poor assembly.

With no “black box” effect, PHOSPHOR® Quality offers all the explainability necessary to justify its decision thanks in particular to a heat map allowing the identification and location of the defect. Self-learning, the model progresses as learning progresses thanks to the feedback that the operator can make on his predictions and adapts to production variations. To reduce the number of false positives, the operator can also teach the machine that a dust defect does not contravene, for example, the quality charter. Our solution, man-machine oriented, has been entirely designed to be 100% collaborative.

Automatically generated by the system, the report lists all control points, together with a heat map showing the potentially defective area and the photo of the control point. On the basis of this history, it is possible, with supporting photos and diagnosis, to demonstrate in the event of a customer dispute that the steering column number 6522, for example, did not present any fault on the front right clip.

WHAT ARE THE MAIN USE CASES ADDRESSED BY PHOSPHOR® Quality?

In the majority of application cases, there is a search for automation. Visual inspection can be particularly painful and repetitive. Operator vigilance also decreases over time. It is not a question of replacing the human, but on the contrary of increasing it by providing it with decision-making assistance and relieving it of difficult tasks.

The automotive industry and its zero-defect policy is a pioneer in the field. For example, our partner ThyssenKrupp is particularly vigilant in detecting defects on its columns. In the medical sector, Septodont has used our solution to control the quality of glass tubes containing anesthetic in dental surgery. The aeronautics, plastics, agrifood and luxury sectors (watchmaking, jewelry, cosmetics) are also concerned.

WHAT ARE YOUR DEVELOPMENT PROSPECTS?

AnotherBrain works to continuously improve its solution and add new features. It is possible to update our software without having to change hardware equipment to take advantage of our rich feature roadmap.

By the end of 2022, PHOSPHOR® Quality will be able to detect part defects in two dimensions, without particular geometry like a frying pan. For this, we will offer a fixed camera lamp, without a robotic arm. In the future, we will also invest in quality control based on the analysis of a sound source such as the sound of a door closing or that of a motor running.

On the commercial level, AnotherBrain intends to capitalize on its success in France before heading internationally. Priority will be given to countries in Europe, particularly in the East, which have maintained visual control while experiencing a shortage of skilled labour.

AnotherBrain wishes to actively participate in the restoration of French industrial and digital sovereignty. Our French technologies perfectly meet the challenges of reindustrialising our territory in a “post-Covid” world, in particular its automation needs. We also make it a point of honor to work with French or European partners.

Proposed content AnotherBrain

Selected for you

ABB partners with Red Hat to strengthen its industrial automation solutions

Leave a Comment