Adaptive UAV Site Surveying for Minimal-Image 3D Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D reconstruction systems for industrial assets rely heavily on manual data collection by human operators, which is error-prone, time-consuming, and computationally expensive, and often requires extensive image datasets that are difficult to process, especially in complex and dynamically changing environments.
Innovation Solution
An autonomous unmanned robot system that plans and adapts its mission in real-time to collect minimal, high-quality data for 3D modeling, using geofences, camera parameters, and asset dimensions to optimize image capture, allowing on-the-fly updates based on sensed local geometry and reducing the need for human intervention and computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If manual data collection by human operators is used, then data can be collected with human judgment and adaptability, but the process becomes error-prone, time-consuming, and requires substantial human intervention
Solution Approach 1:
The system performs self-service through autonomous mission planning and execution. The robot automatically plans its own data collection mission based on asset type and constraints, executes the mission without human intervention, and adapts on-the-fly based on sensed information, eliminating the need for human operators to manually control the data collection process
Solution Approach 2:
The system performs preliminary action by pre-planning the entire data collection mission before execution. The mission planner generates a comprehensive plan considering asset type, dimensions, geofences, and camera parameters in advance, allowing the robot to autonomously execute the pre-planned mission without requiring real-time human decisions
2Reliability
If a large number of images are collected manually, then comprehensive data coverage is achieved, but computational processing becomes expensive and time-consuming
Solution Approach 1:
The system applies partial action by collecting only the necessary amount of data required for accurate 3D reconstruction. The autonomous mission planner calculates the optimal number of images and viewpoints needed based on asset characteristics, avoiding the collection of excessive data that would require lengthy processing while still achieving complete and reliable coverage
Solution Approach 2:
The system performs preliminary filtering and planning to determine exactly which data points are needed before collection begins. The mission planner pre-calculates optimal viewpoints, camera parameters, and sampling rates, ensuring that only relevant data is collected from the start, thereby eliminating the need for extensive post-processing of unnecessary data
3Adaptability or versatility
If manual data collection follows a defined path, then systematic coverage is achieved, but the system cannot adapt to dynamically changing environments or complex asset geometries
Solution Approach 1:
The system implements dynamics through on-the-fly mission adaptation. While the overall mission is pre-planned, the robot continuously senses its environment and asset geometry during execution, automatically adjusting its path and data collection parameters in real-time to adapt to complex geometries and changing conditions without increasing operational complexity for the user
Solution Approach 2:
The system uses feedback mechanisms where the robot continuously senses environmental information and asset geometry during data collection, feeds this information back to the mission planner, and automatically adjusts its mission in real-time. This closed-loop approach enables adaptation to dynamic environments while the automated nature of the feedback process prevents complexity from translating to user burden
4Productivity
If images are collected without focus verification, then data collection speed increases, but reconstruction accuracy degrades due to out-of-focus images
Solution Approach 1:
The system performs preliminary verification of image quality criteria before collection begins. The mission planner pre-calculates optimal camera parameters, focal lengths, and positioning to ensure all collected images will be in focus and meet quality requirements, eliminating the need for post-collection focus verification while maintaining both speed and accuracy
Solution Approach 2:
The system performs self-service quality assurance through automated focus verification integrated into the data collection process. The robot automatically checks image focus quality in real-time and adjusts its position or camera parameters as needed, maintaining high image quality without requiring manual inspection and thereby preserving data collection speed
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for generating a three-dimensional model of an asset includes receiving input parameters corresponding to constraints of a mission plan for operating an unmanned vehicle around an asset, generating the mission plan based on the input parameters including information of a representative asset type, wherein the mission plan includes waypoints identifying locations and orientations of one or more image sensors of the unmanned vehicle, generating a flight path for the unmanned vehicle connecting the waypoints that satisfy one or more predefined criteria, monitoring a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to the next waypoint, determining, at each waypoint, a local geometry of the asset sensed by the one or more image sensors, changing the mission plan on-the-fly based on the local geometry, and capturing images of the asset along waypoints of the changed mission plan.