Using a camera, still and video recordings are collected. Basing on initial learning sessions, the trained system can detect persons in a frame and evaluate their age, emotions, behaviour or movement dynamics (e.g. speed and direction). Having the precise patterns entered as an input, the system can decide if a specific person is allowed to be in their current location or if their behaviour can be treated as unwanted (e.g. the person is influenced by alcohol or narcotics).
Once the system is taught (see Machine Learning section) with precise and detailed patterns of a specific product (a whole variety of fields, from electronic boards, through clothes to meat and fruits), it can use Robotic Senses placed at the manufacturing place to detect any failures, i.e. the items not compliant with the desired standard. Then, depending on the defined process, the system can notify the operator or communicate directly with the manufacturing robot to remove or mark the defective item.
Many mechanical devices (engines, fans, transmission mechanisms) produce sounds during their normal work. The characteristics of a correctly working device can be analyzed and given as a pattern to the system using the Machine Learning techniques. With that knowledge and the sound samples collected with the Robotic Senses while the device is working, the system can detect failures in the mechanical system, or assess its condition (e.g.recognize worn-out gears).