公开(公告)号：US20220066053A1

公开(公告)日：2022-03-03

申请号：US17006833

申请日：2020-08-29

Applicant: Google LLC

Inventor： Stefano Maggiolo ,  Imad Jean Fattouch ,  Andrea Marchesini ,  Zixuan Qu ,  Zifei Tong

IPC: G01S19/51 ,  G01S19/25 ,  G01S19/26 ,  G06K9/62

Abstract: A method of modeling mutable environmental structures includes receiving an estimated location of a Global Navigation Satellite Systems (GNSS) receiver in an environment and a plurality of GNSS-related features corresponding to a GNSS signal received at the GNSS receiver. The method also includes determining a plurality of candidate positions about the estimated location for the GNSS receiver. Each candidate position corresponds to a possible actual location of the GNSS receiver in the environment. For each candidate position, the method further includes generating, as an output from a GNSS localization model configured to receive the plurality of GNSS-related features as input, a respective probability that the respective candidate position includes the actual location of the GNSS receiver. The method also includes selecting from the plurality of candidate positions, the respective candidate position having a greatest probability as the actual location of the GNSS receiver.

公开(公告)号：US11550068B2

公开(公告)日：2023-01-10

申请号：US17006833

申请日：2020-08-29

Applicant: Google LLC

Inventor： Stefano Maggiolo ,  Imad Jean Fattouch ,  Andrea Marchesini ,  Zixuan Qu ,  Zifei Tong

IPC: G01S19/51 ,  G01S19/25 ,  G01S19/26 ,  G06K9/62

Abstract: A method of modeling mutable environmental structures includes receiving an estimated location of a Global Navigation Satellite Systems (GNSS) receiver in an environment and a plurality of GNSS-related features corresponding to a GNSS signal received at the GNSS receiver. The method also includes determining a plurality of candidate positions about the estimated location for the GNSS receiver. Each candidate position corresponds to a possible actual location of the GNSS receiver in the environment. For each candidate position, the method further includes generating, as an output from a GNSS localization model configured to receive the plurality of GNSS-related features as input, a respective probability that the respective candidate position includes the actual location of the GNSS receiver. The method also includes selecting from the plurality of candidate positions, the respective candidate position having a greatest probability as the actual location of the GNSS receiver.