Invention Grant
- Patent Title: Method for calibrating a multi-sensor system using an artificial neural network
-
Application No.: US16879335Application Date: 2020-05-20
-
Publication No.: US11415433B2Publication Date: 2022-08-16
- Inventor: Huang Xiao , Marco Lampacrescia
- Applicant: Robert Bosch GmbH
- Applicant Address: DE Stuttgart
- Assignee: Robert Bosch GmbH
- Current Assignee: Robert Bosch GmbH
- Current Assignee Address: DE Stuttgart
- Agency: Norton Rose Fulbright US LLP
- Agent Gerard Messina
- Priority: EP19176913 20190528
- Main IPC: G01C25/00
- IPC: G01C25/00

Abstract:
A method for calibrating a sensor with respect to a reference trajectory indicating a number of timely successive data values using an encoder/decoder neural network with an encoder function and a successive decoder function. The method includes training the encoder function and the decoder function of the encoder/decoder neural network with respect to a reference trajectory so that a reconstruction error of the reference trajectory at an input of the encoder function and a reconstructed reference trajectory at an output of the decoder function is minimized; training a calibration vector with respect to a sensor trajectory so that a reconstruction error between an output sensor trajectory and a corresponding reference trajectory is minimized, the output sensor trajectory is obtained by applying the calibration vector on the sensor trajectory and the encoder function and decoder function on the the calibrated sensor trajectory; and applying the calibration vector to calibrate the sensor.
Public/Granted literature
- US20200378793A1 METHOD FOR CALIBRATING A MULTI-SENSOR SYSTEM USING AN ARTIFICIAL NEURAL NETWORK Public/Granted day:2020-12-03
Information query