Design and Implementation of a Video-Frame Localization System for a Drifting Camera-Based Sewer Inspection System

Area of Science:

• Civil Engineering
• Robotics
• Computer Vision

Background:

• Sewer pipe inspection is costly and challenging, especially in small-diameter pipes.
• Existing wireless camera systems struggle with accurate localization due to limited Wi-Fi range and lack of GNSS signals underground.
• Identifying damaged pipe areas requires precise location data for each video frame.

Purpose of the Study:

• To develop a cost-effective sewer inspection system using drifting wireless cameras.
• To overcome the localization challenges in small-diameter underground pipes.
• To improve the accuracy of video frame positioning for damage identification.

Main Methods:

• Utilizing drifting wireless cameras to record interior sewer pipe videos.
• Transmitting video data to access points in utility holes for further transfer to a video server.
• Implementing a novel localization function based on linear interpolation using camera-detected landmarks and image processing.
• Conducting experiments in underground sewer pipes to test localization accuracy.

Main Results:

• Successfully detected all utility holes as landmarks during experiments.
• Achieved a maximum location estimation accuracy of less than 11.5% of the maximum landmark interval.
• Demonstrated the feasibility of the proposed landmark-based localization method in underground environments.

Conclusions:

• The proposed landmark-based localization method effectively addresses the limitations of Wi-Fi and GNSS for sewer inspection.
• This system offers a viable, low-cost solution for accurate sewer pipe inspection and damage assessment.
• The improved localization accuracy facilitates better identification of damaged pipe sections, aiding maintenance efforts.