-
公开(公告)号:US20210064980A1
公开(公告)日:2021-03-04
申请号:US16498617
申请日:2018-03-26
发明人: Stefan HEINRICH , Elias STRIGEL , Martin PFITZER , Jan GIEBEL
摘要: A method of ascertaining disparities in sensor data uses at least one neural network implemented in a controller of a vehicle. The method involves capturing (101) a learning data record from temporally successive raw sensor data, evaluating (102) the learning data record to train the neural network exclusively based on the learning data record of the captured raw sensor data, ascertaining (103) expected sensor data, comparing (104) the ascertained expected sensor data with sensor data currently captured by the sensor arrangement, and ascertaining (105) a disparity between the currently captured sensor data and the ascertained expected sensor data.