Single-Camera AR Object Placement via Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing augmented reality systems require multiple cameras and sensors to accurately render virtual objects with depth perspective, making them costly and impractical for widespread adoption in mobile consumer devices.
Innovation Solution
The use of a single rear-facing camera, gyroscope, magnetometer/compass, and location-detection system in mobile devices, combined with a backend system that generates a simplified model of the target image, allows for the placement and presentation of virtual objects within an augmented reality sphere using geographic location, direction, and tilt data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple cameras and sensors are used to accurately render virtual objects with depth perspective, then the rendering accuracy and authenticity are improved, but the device complexity and cost increase
Solution Approach 1:
The patent segments the depth perception function across multiple data sources (single camera image, gyroscope orientation data, magnetometer directional data, location data) rather than requiring multiple cameras. Each sensor type contributes a specific aspect of spatial information that is processed separately and then integrated to achieve accurate depth perspective rendering.
Solution Approach 2:
The patent introduces a backend system as an intermediary that receives data from mobile devices, generates simplified models of target images, and processes the integration of multiple sensor data types. This intermediary handles the complex computation of combining single-camera images with orientation, direction, and location data to produce accurate depth perspective without requiring complex hardware in the mobile device.
2Reliability
If multiple cameras and sensors are used to accurately render virtual objects, then the authenticity of virtual object presentation is improved, but the cost of mobile consumer devices increases
Solution Approach 1:
The patent creates simplified models (copies) of the target images and spatial environments on the backend system. These simplified models capture the essential geometric and spatial characteristics needed for accurate virtual object placement, allowing the system to achieve authentic presentation without requiring expensive multiple-camera hardware in the mobile device. The simplified model serves as a computational surrogate for complex multi-sensor data.
Solution Approach 2:
The patent makes the single camera and associated sensors perform multiple functions: capturing the target image for simplified model generation, providing orientation reference, capturing directional data via magnetometer, and contributing to location-based positioning. This multi-functional use of minimal hardware achieves reliable virtual object presentation without requiring dedicated sensors for each function.
3Ease of manufacture
If a single camera and basic sensors are used, then the device complexity and cost are reduced, but the ability to accurately render virtual objects with depth perspective deteriorates
Solution Approach 1:
The patent compensates for the limitations of a single camera by incorporating data from additional dimensions: temporal (gyroscope orientation over time), directional (magnetometer compass data), and spatial (location data). By integrating information across these different dimensional domains, the system achieves accurate depth perspective rendering despite using only a single camera for visual capture.
Solution Approach 2:
The patent replaces the mechanical/optical depth sensing capability of multiple cameras with a computational approach using basic sensors and backend processing. Instead of using multiple optical sensors to directly measure depth, the system substitutes this mechanical system with computational geometry and trigonometry applied to data from a single camera combined with orientation, direction, and location information.
Data Source
AI summary
Systems described herein allow for placement and presentation of virtual objects using mobile devices with a single camera lens. A device receives, from a first mobile device, a target image captured from a camera and target image data collected contemporaneously with the target image. The target image data includes a geographic location, a direction heading, and a tilt. The device receives, from the first mobile device, a first virtual object definition that includes an object type, a size, and a mobile device orientation for presenting a first virtual object within a video feed. The device generates a simplified model of the target image, and stores the first virtual object definition associated with the target image data and the simplified model of the target image. The device uploads the first virtual object definition and the target image data, so the first virtual object is discoverable by a second mobile device.