Viaposte experiments with artificial intelligence at its Poissy and Garonor ELIs

In partnership with Sick, Viaposte's industrial management has launched an artificial intelligence experiment on its Sort & Drive sorters at its Poissy and Garonor ELIs. This test aims to teach the sorters to automatically detect unstable parcels they are injected. 
The experiment was carried out in several phases.

Phase 1 : Education of the machine

In order for the products to be recognised, the machine first had to be taught to recognise standard and non-standard parcels according to criteria of product type, dimensions and instability on the sorter. 
The teams looked for different non-standard products in order to educate the automaton; this is called machine learning. 
Several product categories were defined such as unstable, round, cylindrical, irregular or ¾ empty bags. 

Phase 2 : Decision-making

This time it was a question of developing decision thresholds via artificial intelligence, enabling the robot to qualify the parcel and exclude it if it is not admissible on the sorter.
Once the trial has been completed and validated at the Poissy and Garonor platforms, artificial intelligence could be deployed at other Viaposte sites to further improve the reliability of sorters and reduce the risk of incidents related to non-standard products.
Viaposte's ambition is to deploy artificial intelligence more widely in order to offer increasingly reliable solutions to its customers.
Discover our offers.

Also visit our social networks