[10:00 Sat,18.March 2023 by Thomas Richter]
Researchers at Carnegie Mellon University have presented an interesting new DeepLearning method that makes it possible to determine the position and posture of multiple people in a room based only on an analysis of WLAN signals. The researchers see their method as a cost-effective and, in some use cases, better solution in terms of quality and data protection compared to systems that use RGB camera, LiDAR or radar for the same task.
For example, using the radiation of existing WLAN routers in a room has the advantage that, in contrast to the use of an RGB camera, the illumination of the room or mutual occlusion of persons is not a problem. In addition, unlike the use of radar and LiDAR technology, no special and expensive hardware is required - the experiment was carried out, for example, with two commercially available TP-Link AC1750 routers worth around 30 euros each. In contrast to the use of a normal camera, room surveillance via WLAN does not produce any image data that may be questionable in terms of data protection in some scenarios.
The team has developed a specially trained DeepLearning AI that maps the phase and amplitude of WiFi signals onto coarse human 3D models with 24 different segments and can thus map different postures (even of several people - successful tests were conducted with up to five). According to the study, the new method can estimate the postures and positions of multiple people similarly well as a solution that uses 3D image analysis, based only on WLAN signals. The prerequisite is three WLAN transmitters and receivers each, but this can also be realized - as in the experiment - in the form of an arrangement of a transmitting and a receiving WLAN router, each with three antennas.
The researchers suggest applications in the field of security technology, among others - using an already existing WLAN, one&s own home could be monitored remotely for illegal intruders, or it could be checked in a retirement home, for example, whether an elderly person has fallen and needs help.
Bild zur Newsmeldung:
deutsche Version dieser Seite: KI verwandelt WLAN Router in Raum-Radar