Title:
ATTITUDE PREDICTION METHOD AND APPARATUS, AND ELECTRONIC DEVICE
Document Type and Number:
WIPO Patent Application WO/2019/192172
Kind Code:
A1
Abstract:
An attitude prediction method and apparatus, and an electronic device. The method comprises: according to a pre-set sampling frequency, sampling attitude data generated by an attitude sensor when a user moves in order to obtain an original data set (S201); according to the original data set, determining the number of layers of a neural network and the number of nodes on each layer, and obtaining a neural network model that has been trained (S202); and during one instance of attitude prediction, inputting the collected attitude data within a pre-determined period of time into the neural network model to obtain attitude data output by the neural network model and corresponding to a target moment after a current moment (S203), wherein the pre-determined period of time is a period of time determined by means of the current moment and a pre-determined moment before the current moment. It can be seen that, in the technical solution, a neural network model trained by means of machine learning is used to perform attitude prediction, such that the efficiency requirementis are met and the computation complexity is also reduced, and a small movement pattern in a historical attitude can also be captured, thus maintaining a high prediction accuracy and improving the experience, such as the immersion, of a user.
Inventors:
ZHU YUGE (CN)
DAI TIANRONG (CN)
CAI LEI (CN)
DAI TIANRONG (CN)
CAI LEI (CN)
Application Number:
PCT/CN2018/112854
Publication Date:
October 10, 2019
Filing Date:
October 31, 2018
Export Citation:
Assignee:
GOERTEK INC (CN)
International Classes:
G06F3/01
Foreign References:
| CN108664122A | 2018-10-16 | |||
| CN106682733A | 2017-05-17 | |||
| CN103984416A | 2014-08-13 | |||
| US20160132773A1 | 2016-05-12 |
Download PDF: