Edge-Computed AR Pose Prediction for Lightweight Wearables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing augmented reality devices face challenges in providing real-time image output while minimizing size and weight, leading to user discomfort due to discrepancies between detected poses and output images.
Innovation Solution
An augmented reality system that distributes computational tasks between a wearable device and a separate virtual image generating device, utilizing inertial and stereo data to predict future poses and generate virtual images efficiently, reducing latency and device size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If computational tasks are performed locally in the augmented reality device, then real-time image output is achieved, but device size and weight increase
Solution Approach 1:
The patent divides the computational system into two segments: a lightweight augmented reality device worn by the user and a separate remote computing device. The AR device performs only lightweight tasks (capturing images, tracking pose, generating depth maps) while heavy computational tasks (generating equirectangular images, rendering virtual content) are performed remotely, resolving the contradiction between real-time processing and device weight.
Solution Approach 2:
The patent introduces a communication network as an intermediary between the AR device and the remote computing device. This intermediary enables real-time data transmission without requiring heavy processing hardware in the wearable device, allowing real-time image output while keeping the device lightweight.
2Volume of moving object
If device size is reduced for user comfort, then processing power is insufficient for real-time image generation
Solution Approach 1:
The patent segments the processing workload between the compact AR device and a remote computing device. The small AR device handles only lightweight sensing and tracking tasks, while the remote device performs computationally intensive virtual image generation, thus achieving both small device size and high productivity.
Solution Approach 2:
The patent replaces the mechanical/computational processing system within the AR device with a remote computing system connected via network. This substitution allows the wearable device to remain small while accessing the processing power of a remote system.
3Ease of operation
If image processing is performed in real-time, then user comfort is improved, but computational load increases device complexity
Solution Approach 1:
The patent segments the system into a simple AR device for data collection and a complex remote device for processing. This segmentation allows real-time processing that improves user comfort while containing computational complexity in a separate system not worn by the user.
Solution Approach 2:
The patent implements a feedback loop where the system continuously tracks user pose, predicts future pose, and adjusts the virtual image accordingly. This self-service mechanism maintains real-time synchronization automatically, improving user comfort without requiring complex manual control systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of generating augmented reality includes obtaining external data, obtaining current pose data corresponding to a position of an augmented reality device, through edge computing, based on the obtained external data, obtaining first future pose data and second future pose data corresponding to movement of the augmented reality device, based on the obtained current pose data, generating a virtual image through the edge computing, based on the obtained first future pose data, and generating a virtual video, based on the obtained second future pose data and the generated virtual image.