Autonomous Track Inspection Drone With GPS-Free Visual Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current railroad track inspection systems rely on human pilots, leading to inefficiencies due to pilot availability, subjective inspection routes, and delayed data processing, which can result in inadequate inspection schedules and reduced accuracy.
Innovation Solution
A fully autonomous drone-based system that uses optical images for navigation and health evaluation of track components, employing computer vision and edge computing for real-time data processing and obstacle avoidance, without relying on GPS, enabling immediate data assessment and optimized flight routes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If human pilots control drone-based track inspection, then operational flexibility and adaptability are maintained, but inspection efficiency and timeliness deteriorate due to pilot availability constraints and subjective route selection
Solution Approach 1:
The drone system performs self-navigation along the track using computer vision to identify rail features and calculate position, eliminating dependence on human pilots for route selection and flight control. The system autonomously processes inspection images and generates reports,实现自我服务式的轨道检测
Solution Approach 2:
The patent replaces human mechanical control with automated computer vision-based navigation. Optical images are processed to extract track features and determine drone position, substituting pilot judgment and manual control with algorithmic image processing and autonomous flight control
2Measurement precision
If image data is transferred to data center for analysis, then comprehensive processing is achieved, but inspection timeliness deteriorates due to data transmission delays
Solution Approach 1:
The patent introduces edge computing capability directly on the drone platform, moving data processing from centralized cloud-based data centers to distributed edge devices. This dimensional shift in computation location enables real-time image analysis while maintaining assessment accuracy, eliminating data transmission delays
Solution Approach 2:
The system performs preliminary data processing and analysis directly on the drone during flight operations. Inspection images are processed in real-time to generate track condition assessments immediately at the inspection location, rather than waiting for post-mission data center analysis
3Measurement precision
If GPS positioning is used for drone navigation, then location accuracy is improved, but system reliability deteriorates in GPS-denied environments
Solution Approach 1:
The patent introduces computer vision-based visual servoing as an intermediary navigation system that works independently of GPS. The system uses optical images to extract track features and calculate drone position relative to the rail, providing reliable navigation in GPS-denied environments through visual feature matching and geometric calculation
Data Source
AI summary
Described herein is a fully autonomous drone-based track inspection system that does not rely on GPS but instead uses optical images taken from the drone to identify the railroad track and navigate the drone to cruise along the track to perform track inspection tasks; track images are taken by the onboard drone camera and processed to provide both navigation information for autonomous drone flight control and track component health evaluation.