- ifm mate
- Technology
ifm mate: The technology
How does ifm mate work?
Industry 4.0 is still firmly focused on the digitalisation of machines and processes, yet people still have to perform four out of five activities in a typical production process – something that goes almost unnoticed by the ever-increasing digitalisation. Does it therefore not make sense for a sensor specialist to use the human hand as a sensor?
The application utilises a combination of object recognition – i.e. hands – and the individual tracking of each recognised hand. A combined 2D/3D camera enables the precise detection of the handling position. In the camera's field of vision, the human hand acts as a sensor to ensure the positions of the hands can be tracked without any further technical or visual aids. Any deviation from the specified process steps results in messages appearing in the live monitor image, providing digital support to the worker to ensure increased safety when executing the work processes.
Deep learning and artificial intelligence open up completely new potentials
New forms of artificial intelligence, such as deep learning approaches, create new opportunities to solve even complex image recognition problems. Object recognition in the industrial environment has always referred to components that look as similar as possible and can thus be easily described as objects from the point of view of image processing. But when it comes to recognising objects that are dynamic and that always look slightly different, conventional algorithmic approaches reach their limits. The human hand is a prime example: Not only do they vary from person to person, but also in shape, pose, rotation and angular representation. And you should not forget hand movements, either.
But that is where deep learning really excels: Neural networks can recognise precisely such things, through high volumes of data that can be learned in advance.