US 12,248,055 B2
Real time object motion state recognition method using millimeter wave radar
Jiun-In Guo, Hsinchu (TW); and Hung-Yu Liu, Hsinchu (TW)
Assigned to NATIONAL YANG MING CHIAO TUNG UNIVERSITY, Hsinchu (TW)
Filed by National YANG MING Chiao Tung University, Hsinchu (TW)
Filed on Apr. 6, 2022, as Appl. No. 17/658,131.
Claims priority of application No. 110144263 (TW), filed on Nov. 26, 2021.
Prior Publication US 2023/0168361 A1, Jun. 1, 2023
Int. Cl. G01S 13/58 (2006.01); G01S 7/40 (2006.01); G01S 7/41 (2006.01)
CPC G01S 13/588 (2013.01) [G01S 7/403 (2021.05); G01S 7/415 (2013.01); G01S 7/417 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method of real-time recognizing an object motion state by using a millimeter wave radar, comprising the following steps:
detecting an object in response to at least one mixed signal by receiving a mixed signal formed by mixing a sweep transmitting signal transmitted to the object and a sweep receiving signal received from the object;
performing a first processing on the at least one mixed signal to obtain M frames, each of which has a first feature information and a second feature information and corresponds to a specific time point, wherein the first feature information includes M ranges, and the second feature information includes M azimuths;
inputting the M frames into a two-dimensional (2D) convolution model to extract a temporal position feature of the object in each of the M frames, wherein the 2D convolution model includes an input parameter including a time parameter; and
inputting the M frames into a two-dimensional (2D) convolution model to extract a temporal position features of the object in each of the M frames, wherein the 2D convolution model includes an input parameter including time parameter; and
performing a second processing on the extracted temporal position features to recognize the object motion state by a voting mechanism.