Environment Model Update for AR Using Change Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating or updating a model of a real-world environment for augmented reality (AR) and virtual reality (VR) applications is computationally expensive, requiring significant processing power and storage, especially on portable devices with limited resources, and outdated models can lead to inaccuracies in AR experiences.
Innovation Solution
A method and apparatus for updating a stored model of an environment based on captured observations, focusing processing resources on regions likely to change, such as objects that are more likely to move, to maintain a relevant and efficient model representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the model is updated frequently to maintain accuracy, then the model remains relevant for AR applications, but processing resources and energy consumption increase significantly
Solution Approach 1:
The patent applies local quality by differentiating update strategies for different regions of the environment model based on their change probabilities. Instead of uniformly updating the entire model, the system identifies specific regions (objects or areas) with high probabilities of change and updates only those regions more frequently, while maintaining other regions with lower update frequencies. This selective approach maintains model accuracy where needed while reducing overall processing resources and energy consumption.
2Reliability
If the model is updated frequently to maintain accuracy, then the model remains relevant for AR applications, but processing resources increase significantly
Solution Approach 1:
The system implements local quality by applying different update frequencies to different regions of the environment model based on their individual change probabilities. High-probability regions receive more frequent updates to maintain accuracy, while low-probability regions are updated less frequently, optimizing the allocation of processing resources across the entire model.
Solution Approach 2:
The patent segments the environment model into multiple regions or objects, each with its own change probability assessment. This segmentation allows the system to independently manage update frequencies for each segment based on its specific characteristics and likelihood of change, rather than treating the entire model as a single unit that requires uniform updating.
3Reliability
If processing focuses on all regions uniformly, then comprehensive model accuracy is maintained, but energy consumption and processing resources are wasted on stable regions
Solution Approach 1:
The patent applies local quality by tailoring the update frequency and processing intensity to the specific characteristics of each region in the environment model. Regions with high change probabilities receive more intensive processing and frequent updates, while regions with low change probabilities receive minimal processing. This eliminates energy waste on stable regions while maintaining accuracy where changes are likely to occur.
Data Source
AI summary
Examples of the present disclosure relate to methods for performing image processing. In one such example, a model of an environment is stored. Image data representative of a captured observation of the environment is obtained and the image data is processed using an object classifier to determine whether one or more objects belonging to a plurality of predetermined classes of objects is present in the environment. The stored model is updated based on the processing. In response to determining at least one of the one or more objects is present, object class characteristic data indicative of a characteristic of the at least one object that is determined to be present is retrieved and a probability of change associated with the at least one object is determined on the basis of the object class characteristic data. The updating of the stored model is adjusted based on the determined probability of change.