MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202514121883 A) filed by Robert Bosch Gmbh on December 04, 2025, for Keypoint Detection For Improved Object Detection In A Radar System.
Inventors include Braun, Sascha; Bartler, Alexander; Jenet, Daniel Kenta; Armanious, Karim Adel Dawood; Schreiber, Marcel; Reese, Marco; Quach, Maurice Number; Ulrich, Michael; and Ziegler, Patrick.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: The present subject matter relates to a method for object detection in a radar system (2.1, 2.2), comprising providing radar measurements as a list of points, wherein each point includes associated features; creating object hypotheses based on the radar measurements using a trained artificial neural network (1), comprising the steps of creating a feature map (20) based on the radar measurements, creating (S3.1) a probability distribution (22) over the feature map, wherein each probability maximum corresponds to the most probable key point (24.1, 24.2) of an object, regressing (S3.2) each key point (24.1, 24.2), filtering the probability maxima, and outputting an object hypothesis for each key point (24.1, 24.2). The present subject matter further relates to a radar system (2.1, 2.2) for performing such a method.
Disclaimer: Curated by HT Syndication.
