Desarrollo de un sistema prototipo basado en visión artificial para la detección y alerta de posturas forzadas a nivel del tronco en los puestos de trabajo de cajeras en los supermercados

Abstract

In this degree research, a prototype system based on artificial vision was developed in order to detect and alert forced postures at the trunk level in people working as cashiers in supermarkets for which the detection is based on the parameters established by the standard NTE INEN-ISO 11226 of the trunk posture whose coding was developed in Python software using the OpenCv and MediaPipe Pose libraries, for the alert, an electronic circuit was built based on an ESP32 module that communicates Python with Arduino through the use of UDP sockets and a vibrating micro motor that is activated when there is a forced posture and incites it to take corrective actions. Once the prototype was completed, it was validated using AutoCAD software, where a relative error of 0. 2% considering the trunk inclination angle of AutoCAD as the exact measurement and the value obtained by the prototype as the experimental measurement. In addition, STT-IWS inertial sensors were used in conjunction with the ¡Sen software to evaluate the trunk posture of the prototype and the aforementioned software, yielding reliable results, Then the operation tests were performed in a supermarket where the person adopted different angles of inclination of the trunk based on the specified standard were recommended and not recommended, it allowed to improve ergonomically the workstation showing the operability of the prototype. It is concluded that the prototype has a very wide application not limited to supermarket cashiers but also to jobs involving static work postures and it is recommended in future research to complete the study with the evaluation of all parts of the human body.