Autonomous 3D Asset Inspection With Edge-Based Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In-person inspections of manufactured articles and facilities are time-consuming, prone to human error, and costly, with limited data capture and analysis capabilities, especially when comparing multiple inspections over time, and existing digital methods face challenges with large data volumes and communication costs.
Innovation Solution
A system and method for autonomous inspection using sensor apparatuses with multiple sensors that capture and process three-dimensional data, utilizing edge computing to reduce data size and enable real-time comparison and visualization of object similarities and differences across inspections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If in-person inspections are performed by inspectors using various senses and specialized metrology equipment, then detailed observations can be obtained, but the process becomes time-consuming and costly
Solution Approach 1:
The patent replaces human inspectors and their physiological observation systems with automated sensor apparatuses that capture data electronically. Multiple sensors (cameras, LIDAR, thermal sensors) automatically record measurements without requiring human traversal of the site, thereby maintaining measurement precision while dramatically reducing inspection time and cost.
Solution Approach 2:
The system creates digital copies (three-dimensional models) of the inspected objects and environments. These digital models preserve all measurement data in a compact format that can be stored, analyzed, and compared without requiring physical re-inspection, thus eliminating time loss while maintaining observation accuracy.
2Loss of information
If multiple sensors capture comprehensive data from various positions and angles, then complete object information is obtained, but data volume increases significantly
Solution Approach 1:
The patent segments the comprehensive inspection data into distinct three-dimensional models for each inspected object. Each model contains only the data relevant to that specific object from multiple sensor inputs, separating complete information from redundant data and reducing overall data volume while maintaining data completeness for each object.
Solution Approach 2:
The system transforms raw sensor data from multiple positions and angles into three-dimensional spatial models. This dimensional transformation organizes comprehensive data into a compact spatial representation that preserves all object information while reducing data volume through efficient spatial encoding.
3Loss of information
If detailed measurements and observations are recorded for every object and inspection point, then comprehensive inspection data is available, but organizing and recalling specific data points becomes complex and time-consuming
Solution Approach 1:
The patent creates structured digital copies of inspected objects in three-dimensional model format. Each model inherently organizes all measurement data by spatial location and object identity, allowing instant retrieval of specific data points through model navigation rather than searching through unorganized datasets. This maintains full data detail while eliminating organizational complexity.
4Loss of information
If large quantities of inspection data are transmitted over networks for processing and analysis, then comprehensive analysis is possible, but communication costs increase and transmission speed is limited
Solution Approach 1:
The patent segments inspection data into discrete three-dimensional models that can be processed independently. Only essential analysis results and anomaly detections need to be transmitted over the network, rather than complete raw datasets. This maintains analysis completeness for local processing while minimizing communication costs and bandwidth requirements.
Solution Approach 2:
The system uses three-dimensional models as intermediary representations between data collection and analysis. These models serve as compact intermediaries that preserve all necessary information for comprehensive analysis while requiring minimal transmission bandwidth, thereby reducing communication costs while maintaining analysis completeness.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A method for performing an autonomous inspection. The method comprises traversing, by an autonomous sensor apparatus, a path through a site having three-dimensional objects located therein. The site includes three-dimensional objects located therein. The method comprises obtaining, by a plurality of sensors on-board the autonomous sensor apparatus, one or more data sets throughout the path. Each of the one or more data sets are associated with an attribute of one or more three-dimensional objects. The method comprises generating, by the first, second, or third processor, a working model from a collocated data set; and comparing, by the first, second, or third processor, the working model with one or more pre-existing models; to determine the presence and/or absence of anomalies. The presence and/or absence of anomalies are communicated as human-readable instructions.