Robot for Mouth Swab Sampling Could Help Medical Staff
Until we come up with something less unpleasant, all the most reliable ways to find out whether you are carrying even the tiniest particle of COVID-19 that exist right now involve unpleasant mucosal sampling procedures. Hardly anyone enjoys the PCR testing process.
Whether it is a procedure through the nose or deep into the throat, to the point where you involuntarily try to push the nurse away, collecting samples of disgusting mucus is not pleasant for patients and probably even less pleasant for medical workers, especially when there is a real chance that the slime you are up to your elbows in carries COVID-19.
Taking an oropharyngeal swab is dreary, messy and dangerous all at once, which makes it an ideal task for a robot, as long as that robot can avoid any forceful contact with the human at that moment.
In an article published in IEEE Robotics and Automation Letters, researchers from the Shenzhen Institute of Artificial Intelligence and Robotics for Society presented a robot designed to protect the safety of medical staff during the sampling process, and to free up their time for more useful and less grueling work. Let's face it, nobody enjoys taking these samples.
"The COVID-19 outbreak has caused mortality and morbidity worldwide. Oropharyngeal swab sampling is widely used for diagnosing COVID-19 in countries. To avoid exposing medical staff to the virus, we developed a robot with a rigid-and-flexible coupling (RFC) with 9 degrees of freedom (DOF) that helps collect swab samples. This robot consists of a vision system, a UR5 manipulator, a micro-pneumatic actuator and a force measurement system," the study says.
The researchers note that the human throat is fragile and easily injured. End effectors built on a rigid housing structure can cause physical damage to the mouth if mishandled. To make things a little less unsafe, this system combines a UR5 arm with a disposable soft micro-pneumatic actuator, connected to a force sensor to actually perform the swab in the throat, with a vision sensor to determine where to press.

The video shows that excessive force on the actuator will cause the robot to stop moving, and that it has no trouble safely collecting samples from real human volunteers.
A concern with this system is that, by design, the patient's head appears to be within the workspace of the UR5, or at least within reach of the rigid components attached to the arm.
Behind this deformable actuator stands a non-deformable linear actuator, supported by a powerful robotic arm with joints that can move at 180 degrees per second. Yes, force detection is included and should stop the arm if it hits something, but that's software, right?

Patients would feel better if the system were designed so that the arm, linear actuator and all other rigid components were positioned so they physically could not cause harm, for example by arranging things so that the person can adjust how far they are from the system, rather than the system having to adjust to reach the patient and stop short of going beyond them.
The researchers acknowledge that "mechanical (constrained) implementation is generally safer for open minimally invasive surgery," but they believe software safety is cheaper and more convenient. It's just that it may be less safe.
In fairness, the first step here is to make sure the system can reliably collect samples, because if it cannot, why worry about anything else, right? PCR testing (using a reference gene rather than COVID) to verify the effectiveness of the sampling method showed that samples taken by the robot were fully qualified on par with samples taken by a professional medical worker. Next, the researchers plan to "improve the reliability of the system and move on to clinical trials."

The robot is expected to reduce risk and free medical staff from long, repetitive sampling work. Compared with a rigid sampling robot, the force-sensitive device developed by the researchers may simplify swab sampling procedures in a safer and gentler way. In addition, for motion planning of the redundant manipulator with nine degrees of freedom, a neural network optimization method with zeroing variable parameters is also proposed.
The developed robotic system was verified by taking swabs on both oral cavity phantoms and volunteers. The average swab collection time for phantoms and volunteers is 18 and 20 seconds, respectively. To maintain sterility of the arm and effector between patients, the micro-pneumatic actuator is made as a disposable device with a quick-release connector, which is easily replaced at the point where the disposable protective film is attached.