Edge 3D Avatar Caching for Low-Latency Telepresence Rendering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Augmented reality telepresence applications using edge computing face inefficiencies in computing power consumption due to rendering 3D Avatar images for repetitive user actions, leading to high latency and increased network costs.

Innovation Solution

A method and apparatus for providing a 3D character image by storing previously generated data in a cache and transmitting cached images based on key and view point similarities to reduce computing resources and processing time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If 3D Avatar images are rendered for all user actions and movements in real-time, then the quality and performance of augmented reality telepresence is maintained, but computing power is inefficiently consumed and processing time increases

Engineering Contradiction:
Improvequality of 3D Avatar renderingVSAvoidcomputing efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system pre-renders 3D Avatar images for predicted user actions and movements before they actually occur. By anticipating future user behaviors based on current state and historical patterns, the system prepares rendering results in advance, so when the actual action occurs, the pre-computed image can be immediately used without real-time computation delay.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system dynamically adjusts rendering parameters such as image resolution, detail level, and computational complexity based on the importance and predictability of user actions. For repetitive or predictable actions, lower-resolution or simplified versions are used, while for novel or important actions, full-quality rendering is performed. This adaptive parameter adjustment optimizes the balance between quality and computing efficiency.

Inventive Principle:
Principle #35Parameter changes

2Manufacturing precision

If 3D Avatar images are rendered with high quality for all movements, then the visual fidelity is improved, but the amount of data transmitted through the network increases

Engineering Contradiction:
Improvevisual fidelity of 3D AvatarVSAvoidnetwork data transmission volume
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The system applies different quality levels to different regions or aspects of the 3D Avatar rendering. For example, the face and upper body which are most visible and important for communication receive higher resolution and detail, while less important regions use lower quality. This selective quality allocation maintains visual fidelity where it matters most while reducing overall data transmission requirements.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system transmits only the necessary portions of the 3D Avatar data at full quality, rather than transmitting complete high-resolution models for all actions. For repetitive or minor movements, partial updates or lower-fidelity representations are transmitted, reserving full-quality transmission for significant or novel actions that require high visual fidelity.

Inventive Principle:
Principle #16Partial or excessive action

3Speed

If real-time rendering is performed for every user action, then the responsiveness of the system is maintained, but the processing time and latency increase

Engineering Contradiction:
Improvesystem responsivenessVSAvoidprocessing time
Core Design Contradiction:
SpeedVSLoss of time

Solution Approach 1:

The system performs rendering operations in advance for predicted user actions, storing the results in a cache or buffer. When a user action occurs, the system checks whether a pre-rendered image is available and uses it immediately if applicable, avoiding real-time computation and reducing latency. This lookahead approach maintains responsiveness while eliminating unnecessary processing time.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system combines multiple rendering operations into batch processing where possible, or merges the rendering pipeline with prediction and caching operations. By coordinating these functions together rather than executing them sequentially, the system reduces overall processing time and improves responsiveness to user actions.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP4167188B1Edge data network for providing 3D character image to terminal and operation method therefor
Publication Date: 2025.12.24 SAMSUNG ELECTRONICS CO LTD
  • EP4167188B1 patent drawingFigure 1
  • EP4167188B1 patent drawingFigure 2
  • EP4167188B1 patent drawingFigure 3

AI summary

An edge data network for providing a three-dimensional (3D) character image to a user equipment and an operating method thereof are provided. The edge data network obtains key points information including feature point coordinates related to the body parts of a first user, from a first user equipment via a network, and obtains view points information including virtual position coordinate value information of virtual view points from which a second user views a 3D character image from a second user equipment, measures a key points similarity and a view points similarity by respectively comparing the obtained key points information and view points information with key points information and view points information cached in a data cache, and reads out a 3D character image cached in the data cache based on the measured key points similarity and the measured view points similarity, and transmits the read out 3D character image to the second user equipment.