US 12,307,761 B2
Scene-adaptive radar
Lorenzo Servadei, Munich (DE); Michael Stephan, Neubiberg (DE); and Avik Santra, Munich (DE)
Assigned to Infineon Technologies AG, Neubiberg (DE)
Filed by Infineon Technologies AG, Neubiberg (DE)
Filed on Aug. 6, 2021, as Appl. No. 17/396,032.
Prior Publication US 2023/0040007 A1, Feb. 9, 2023
Int. Cl. G01S 13/89 (2006.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06V 20/17 (2022.01); G06V 20/10 (2022.01)
CPC G06V 20/17 (2022.01) [G01S 13/89 (2013.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06V 20/194 (2022.01)] 26 Claims
OG exemplary drawing
 
1. A method comprising:
receiving first radar data from a millimeter-wave radar sensor;
receiving a set of hyperparameters with a radar processing chain, wherein the hyperparameters are configuration parameters governing behavior and performance of the radar processing chain;
generating a first radar processing output using the radar processing chain based on the first radar data and the set of hyperparameters;
updating the set of hyperparameters based on the first radar processing output using a hyperparameter selection neural network;
receiving second radar data from the millimeter-wave radar sensor;
generating a second radar processing output using the radar processing chain based on the second radar data and the updated set of hyperparameters;
training the hyperparameter selection neural network using imitation learning;
after training the hyperparameter selection neural network using imitation learning, training the hyperparameter selection neural network using reinforcement learning; and
using a reward function based on missed detections, false alarms, and error distances between predicted and actual target locations during reinforcement learning.