Automatic Functional Tests of Cash Registers

The paper presents formal requirements and a prototype of a stand for automatic functional tests of cash registers. The authors are unaware of any automatic cash register test stands offered on the market and operating. An OCR system was used to detect digits and inscriptions on keys, and their spatial coordinates are determined using stereo vision. Deep neural networks were used to detect pictograms from outside the OCR database. Due to the different size of the cash registers, the stereo vision system must be reconfigured in terms of the height of the cameras and their base. Fast stereo calibration based on only a pair of images was used, which provides sufficient measurement properties. The calculated key coordinates are sent to an industrial robot equipped with a flexible pointer adapted to pressing the keys. The key sequence enabling individual tests is entered by the operator based on the cash register documentation. The advantage of the proposed stand is its versatility, which allows for testing of cash registers of different manufacturer and other devices with similar functionality. The stand is intended for implementation at Central Office of Measures.