Gait recognition could be better than face biometrics to ID coboting workers
Citing the need for a workplace robot to know the humans working around it, researchers writing in the journal Scientific Reports say gait recognition is the better option compared to face biometrics.
Scientists at the University of León, in Spain, created an application called Brittany, a tortured acronym for Biometric Recognition Through Gait Analysis. Brittany identifies people using behavioral biometrics that analyze “the features of a specific action performed by a person.”
Brittany, using a custom model and tested in a simulated home-care environment, was 88 percent accurate. A similar project was published by the university in 2019.
Using conventional cameras (capturing depth and color data) mounted on robotics for confirming someone’s identification means loading a lot of code and processing power into machinery typically designed to be more drone-like, more of an electronic commodity.
That or a robot has to be leashed physically or over wi-fi to a remote processor, sacrificing speed of work, according to a paper describing the innovation.
Facial recognition also means having to see at least a significant portion of a person’s face.
Brittany, loaded onto Robotnik’s Orbi-One mobile service robot for the research project, has a single two-dimension lidar to analyze gaits unobtrusively and in real time.
Onboard, Brittany has a convolutional neural network-based tool devised at León University called People Tracking, or PeTra, which builds a map from data collected by lidar reflections. Movement represented on the map is compared to previously collected biometric information about cobot colleagues.
behavioral biometrics | biometrics | biometrics research | gait recognition | identity verification | robots