Fall-down early warning apparatus and method for virtual reality experience

Abstract

The invention discloses a fall-down early warning apparatus and method for virtual reality experience, and mainly relates to the fields of machine learning and virtual reality. A pose classification program capable of identifying a pose action of a user is obtained by training through a machine learning algorithm; eigenvalues are extracted from a real-time pose of the user to serve as input data of the classification program; an output result is a real-time fall-down risk probability of the user; and when a fall-down risk is excessively high or fall-down already occurs, the apparatus gives analarm. The eigenvalues are processed by using a depth sensor with a skeleton information extraction function; 20 joint points taking a main effect during motion of a human body are extracted; and spatial vectors of the eigenvalues are selected by taking a gravity center point of the human body as an origin of coordinates, taking a vertical direction as a Z axis, taking a direction faced by the user as a Y axis and taking an arm stretching direction as an X axis. The spatial vectors of the 20 joint points of each frame of image serve as a group of eigenvalues. After the feature extraction is finished, eigenvalue data is input to the classification program, so that the fall-down risk probability can be obtained.

Description

technical field [0001] The invention belongs to the fields of virtual reality (VR for short) and machine learning (ML for short), and in particular relates to a fall prediction method in virtual reality experience and an implementation method using machine learning technology. Background technique [0002] With the continuous development of virtual reality technology and the continuous reduction of equipment costs, virtual reality has gradually entered people's study, work and entertainment. However, due to the fact that users cannot observe the surrounding real environment in the immersive virtual reality experience, coupled with the existence of virtual reality vertigo, not only the comfort of people in the immersive virtual reality experience is greatly reduced, but in severe cases, it will also be caused by dizziness and spatial perception. Misalignment causes users to fall and cause safety accidents. Therefore, there is an urgent need for a method that can effectively ...

AI Technical Summary

Problems solved by technology

The main disadvantages are as follows: (1) In practical applications, if the user's activity area changes (such as item movement), the protection mechanism will fail

(2) In practical applications, users may face various safety hazards (such as collisions and falls), but the current method is too narrow to apply and can only be used for collision warning or fall warning for a single user group

(3) The method based on preset dangerous areas has great contingency. The so-called "safety" and "dangerous" are preset and cannot respond to real-time environmental changes. It is only applicable in scenarios where the accuracy of early warning is not high.

Images