Sensing a coughing person using a depth camera and microphones for effective infection control in indoor environments

Airborne infectious diseases have been a threat to human health for decades. Among various infection control methods, source control is the most direct and effective. To implement source control measures, it is crucial to accurately identify the index persons and their locations in the indoor environment. Coughing is one of the indicators that a person may have acquired a respiratory infection, although healthy persons may also cough. Therefore, this study developed a method based on depth camera and microphone sensors to detect cough events and identify the location of the corresponding coughing person. The depth camera was employed to achieve human recognition and obtain the coordinates of persons in the indoor environment. Microphone sensors were used to collect audio samples, enabling the detection of cough events and the identification of the location where the cough occurs based on machine learning. Experiments were conducted to validate the proposed method. The results show that the proposed method successfully identified all cough events and accurately localized the coughing person with an error of 604 mm or less. Finally, the proposed method was applied in the source control of person-to-person particle transport in a laboratory indoor environment with a robot equipped with an air cleaner to demonstrate the value of this method for infection control. The results show that, compared with the air cleaner off and the stationary air cleaner on cases, the air-cleaning robot with the proposed coughing person identification and localization method reduced the receptors’ exposure to exhaled particles by up to 75.8%. • A method with depth camera and microphones was developed for locating a coughing person. • The coughing person localization model was validated by experiments. • The proposed method was applied in the source control of person-to-person particle transport.