Use case
Many accidents happen when safety devices are disconnected “for maintenance”, when a box is left in the passage, when raw material is stacked in a disorderly way, when a still hot part is not secured, when a surface treatment bath is left without supervision…
In each of these situations, it is possible to teach a neural network to recognize risks from examples. For this purpose, for example, pictures representing various normal and risky situations are provided to the neural network. Other non-visual data can be integrated, such as the operating status of a machine. The neural network learns from these situations and alerts you if it believes that it is in a similar situation to a risky one it has already encountered.
The goal of this is not to replace good old-fashioned light curtains or sensors, but to double down on safety by having an additional control that cannot be disconnected and is able to alert you to potential hazards to prevent accidents.
Other more specialized and programmatic approaches can be deployed to provide solutions to specific issues.