Self-Checkout Retail
Main advantages of Self-Checkout systems (SCO)
Self-Checkout (SCO) systems help the retailer to face the many challenges of the current situation of competition and demands from the administration:
- Increase the efficiency of operations
- Reduce labor cost
- Retrench of space
- Improve job availability
- Increase customer satisfaction
- Grow and not lose sales
- Address legal requirements
- Improving the perception of hygiene in critical moments such as COVID
SCO's current systems combine the agility of a point of sale with highly advanced security features, using artificial intelligence and artificial vision algorithms to control the various security parameters in the purchase process.
As for the point of sale, the SCOs reproduce all the classic functionality of the point of sale, which, being already a very mature product, is difficult to optimize:
- Barcode or RFID reading
- Possibility of pick list (the client manually selects his product)
- Reading of barcodes of section scales and ticket capture
- Barcode reading with weight/unit information
- Application of promotions
- Loyalty and card balance management
- Collection management in various payment methods and associated with different peripherals
Technological innovation for fraud control
So far there is not much news regarding what is usually used in any business. Where the most innovative and rapidly evolving technologies are applied is in fraud control.
The great advantage of a SCO is the saving of labor and the reduction of queues thanks to offering alternatives to the customer to go through the classic checkout. So far so good, but this great advantage provided by any SCO becomes a security challenge, since the fact of being an unattended system implies a greater possibility of fraud.
The classic method of control by weight, although it is the most widespread, has several chronic problems:
- The ERP must be fed with the theoretical weights and tolerance margin for all the items or families that you want to control
- Any random weight variation (a promotion, a slight change in packaging, etc.) can imply a false positive from which the system does not learn.
- It involves many manual interventions by the checkout staff, which hinders the classic sale, sometimes reaching 20-30% of the sales of certain items.
- The classic fraud of “not reading” an article, or reading one of the same weight and taking another, continues to rely too heavily on traditional surveillance by SCO staff, reducing labor savings considerably.
To optimize security and the management of staff time, we will detail as follows some interesting improvements that we are already applying to our SCO projects with Partner-Tech technology and our own AV algorithms.
Weight control on SCO scales with AI
Applying artificial intelligence algorithms with Self-Learning, the system learns the weights of each reference. By automatically acquiring data, you can establish the weight tolerance ranges per item based on a simple function that calculates the Standard Deviation and defines the tolerance that covers 99% of the measurements (3σ), however, the staff can manually force a weight change based on a new format for future measurements of a certain product
We illustrate with the typical example of whiskey and potatoes from Partner-Tech's Crime Predictor:
Control of hand movements and validation of the scanning/selection of articles by the client
This technology is the most recent, and we are starting the first prototypes with the algorithms in development. These AV algorithms follow the customer's hands and validate that the product's barcode is read or selected from the pick-list on the screen.
These algorithms, which are in the final stages of development, will soon make it possible to control that all the items in the basket are read, enhancing security control by the staff and minimizing the hours of supervision by the store's security personnel.