A radar four gestures: hardware and software platforms that use millimeter waves to complete the fine gesture recognition

Anfuli super cool gesture identifies that the radar is fresh, a radar identifies four gestures. Compared to the camera, millimeter wave radar can better protect customer privacy, and the algorithm does not require GPU and neural network support, so the reaction is faster. In response to bad weather, millimeter wavelet has natural advantages, all-day sensors, dust rain fog can make the camera fail, can not access effective pictures, and powerful nerve networks can not be correctly identified, millimeter wave radar is strong due to penetration It can still work effectively in harsh environments. We can see that the program of laboratory personnel in the area millimeter-wave radar can be spin-wave radar to detect correctly waved from left to right and right to left waving; rotating clockwise rotation and gesture recognition, including counter-clockwise, and extremely subtle finger rubbing can be properly detected. 2016 Google demonstrated stunning radar chip --Project Soli. To capture the subtle gestures based on 60GHz FMCW radar, to achieve control of the hardware. In the display video, an email can be implemented, which can replace the touch screen cross-air operation control mobile phone, which can replace the touch board operating the computer, and even the developer has a function of implying the typhoon! Compared to the current popular two-dimensional touch screen system, its control gestures are three-dimensional stereo, so it seems more intuitive, more vivid, coupled with the "electromagnetic wave", which only disseminated, quite a magical effect . Nowadays, this development platform is no longer unable to make ordinary developers. Anfuli Beijing RF Lab has developed a hardware and software algorithm based on English Briting Company BGT60TR13 chip, which is easy to develop its own Product, please refer to the presentation video of the platform. Hardware platform Infineon 60GHz radar chip BGT60TR13,5GHz the sweep bandwidth, the FMCW mode of operation, range resolution can be realized 3cm of fine gestures can be identified. Afri's current gesture recognition algorithm runs on a 60 GHz millimeter Prada platform, sweeping bandwidth is 3GHz, which has been rotated, waving around, waving up and down and squatting fingers. The program currently has an assessment kit available for customer assessment, suite BGT60TR13 evaluation board and SDK running on the PC, as shown in Figure 1. Be Figure 1 60GHz mm wave radar function Software Process To facilitate custom software development, Beijing Avnet laboratory has developed a set of software code, the specific features shown in Figure 2, comprising, 2-D FFT transform, a peak search based hardware acceleration, constant false alarm potential target filter algorithm, based on comparison of the phase difference calculation method of the angle of arrival angle direction, and a gesture in the SDK environment classification, and finally the gesture recognition result on the PC to print. Be 2 a functional block diagram of software Be This article is reproduced from: Avnet