Automatic throat swab collection system

Abstract

Contrary to the issues of low sampling efficiency and high risk of cross-infection with the traditional method of throat swab collection, this study researched on automatic throat swab collection system. In order to collect throat swabs, it was necessary to identify and locate human face and oral information, which is why machine vision was used for dynamic navigation and positioning. Firstly, a deep learning approach was employed to identify the external contour of the oral cavity, enabling the localization of the collection area. Subsequently, real-time depth imaging was utilized to calculate the spatial coordinate information of the target area. Additionally, force control feedback was implemented at the end of the collection to ensure the safety by controlling the contact force the throat swab exerting on the human pharynx. Finally, a robotic arm was used as an executive agency to transport the pharyngeal swab to the designated target area for collection. The experimental results demonstrated that the automatic throat swab collection system successfully executed the planned path and obtained valid throat swab samples.