When we deal with a huge amount of data coming from the Robotic Senses (e.g. camera images, infrared videos, sound recordings), to process them effectively, the Machine Learning comes into play. This means that the data is evaluated based on statistical models to detect patterns and rules and then the “digital brain” of the system is trained with the outcome of that process. After the learning phase, the system can not only make correct decisions basing on the “digital experience” but also refine the rules, to make them even more accurate.
The following examples present the fields where Machine Learning can be effectively applied:
- In the collection of digital microscope videos, the system can identify bacterial species and recognize if the specific organisms are alive.
- The system can evaluate the age and mood of a person on a photograph.
- Basing on the input from a camera placed above the conveyor belt, the system can identify a damaged package and order a robot to take it out from the belt.
- Having the knowledge of what is the sound of a correctly working mechanical device (e.g. a fan rotor), the system can evaluate the sound coming from a sensor and detect a failure of the device.