Mobile Victim Signs Monitoring Through Non-invasive Robotic System

Robotic assistance for search and rescue (SAR) tasks in post-disaster environments focuses on offering faster response times for detecting potential victims and acquiring relevant information for prompt medical evaluation. An important potential feature of these robotic systems is the ability to rapidly and efficiently access victims’ vital signs, such as heart rate, respiratory rate, or body temperature. Current methods are still based on manual inhalation - exhalation counts in a minute to obtain the breathing rate. From the approach of mobile robotics, this work proposes a robotic system for analyzing the respiratory rate condition of victims in SAR missions. The system is composed of a legged-manipulator robot equipped with a RGB sensor at the end effector, which, through convolutional neural networks (CNN), identifies a victim and defines a close relative position, then deploys the manipulator to a pose near the victim’s thoracic region, performs motion data acquisition through optical flow techinque, and process the data to estimate respiratory rate. The system has been tested in indoor and outdoor environments. System evidences robustness on difficult lighting conditions and different patterns on victims’ clothes.