AR Device Positioning via Point Cloud Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for spatial positioning, such as inertial measurement units and GPS, are unreliable in indoor environments, leading to errors and resource-intensive continuous field model updates in industrial applications, while SLAM may cause map data errors when personnel move between areas.
Innovation Solution
A positioning method combining a depth sensor and an inertial measurement unit, using point cloud registration to calculate an updated position and direction by matching a read point cloud with a partial environment point cloud, reducing redundant computations and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If SLAM is used to acquire spatial position and attitude in indoor fields, then positioning capability is improved, but computing resources are excessively consumed and map data errors occur
Solution Approach 1:
The patent divides the complete field model into multiple local area models based on physical barriers and spatial partitions. Each local area model contains only the point cloud data and features relevant to that specific region, eliminating the need to process entire building models. This segmentation reduces computing resources while maintaining positioning reliability in each local area.
Solution Approach 2:
The patent creates different levels of detail in point cloud data based on location importance. Frequently visited areas have higher detail and more features, while less important areas have reduced detail. This local quality differentiation optimizes computing resources by focusing processing power on critical regions rather than uniformly processing all spaces.
2Measurement precision
If continuous field model updating is performed in fixed space fields, then positioning accuracy is maintained, but computing resources are redundantly consumed
Solution Approach 1:
The patent implements periodic re-registration of point cloud data at scheduled intervals or upon triggering events (e.g., significant environmental changes detected by sensors). This periodic action maintains positioning accuracy without requiring continuous model updates, significantly reducing redundant computing operations while preserving measurement precision.
Solution Approach 2:
The system uses feedback from sensor data and positioning residuals to determine when re-registration is necessary. By monitoring changes in the environment and positioning accuracy metrics, the system triggers updates only when needed, avoiding unnecessary computations while maintaining required accuracy levels.
3Speed
If inertial measurement unit is used for spatial positioning, then initial position estimation is obtained quickly, but integral drift occurs and location information is lacking in long-term usage
Solution Approach 1:
The patent merges inertial measurement unit data with point cloud registration results in a hybrid positioning system. The IMU provides quick initial position estimates and continuous tracking, while periodic point cloud registration corrects cumulative drift and provides absolute position references. This combination maintains both speed and accuracy over extended periods.
Solution Approach 2:
The point cloud registration system acts as an intermediary correction mechanism for IMU drift. The registered point cloud data serves as a reference framework that periodically recalibrates the IMU's accumulated position estimates, eliminating integral drift while preserving the speed advantages of inertial navigation.
Data Source
AI summary
A positioning method is provided for an electrical device including a depth sensor and an inertial measurement unit (IMU). The positioning method includes: calculating an initial position and an initial direction of the electrical device according to signals of the IMU; obtaining an environment point cloud from a database, and obtaining a partial environment point cloud according to the initial position and the initial direction; obtaining a read point cloud by the depth sensor; and matching the read point cloud and the partial environment point cloud to calculate an updated position and an updated direction of the electrical device.