Hardware
- PX4-based autonomous drone platform
- Embedded flight-control systems
- Vision-processing systems
- Inspection camera systems
- Autonomous navigation hardware
Robotics
Autonomous Drone System for Copper Cable Recovery and Infrastructure Inspection
Developed an autonomous PX4-based drone platform for utility-infrastructure inspection and copper-cable recovery using Mask R-CNN segmentation, monocular metric depth estimation, and vision-guided autonomous positioning.
Summary
Engineering context
Developed an autonomous PX4-based drone platform for utility-infrastructure inspection and copper-cable recovery using Mask R-CNN segmentation, monocular metric depth estimation, and vision-guided autonomous positioning.
- Category
- Robotics
- Year
- Mar 2025 - Present
- Status
- Research Prototype
- Context
- University of Regina MSc / NSERC IRC (Mar 2025 - Present), in collaboration with SaskTel
My Role
Autonomous Systems and Computer Vision Engineer
Technical Stack
System Architecture
- PX4 autonomous drone platform provided infrastructure-inspection mobility
- Mask R-CNN segmentation identified cable bundles and cutting locations
- Depth Anything V2 estimated monocular metric depth from visual imagery
- Computer vision systems guided autonomous positioning and interaction
- Vision-based drone control demonstrated perception-guided motion behavior
- Autonomous infrastructure-inspection pipeline combined perception and robotic positioning
Engineering Challenges
- Detecting cutting points on cable bundles reliably
- Estimating metric depth from monocular imagery
- Integrating AI perception systems with autonomous drone control
- Stabilizing vision-guided autonomous positioning
- Developing infrastructure-inspection workflows for aerial robotics
Hardware / Firmware / Software
Firmware
- PX4 flight-control firmware
- Embedded flight-system configuration
- Autonomous navigation integration
Software
- Mask R-CNN segmentation pipeline
- Monocular metric depth-estimation system
- Computer vision positioning algorithms
- Autonomous inspection software
- Vision-guided control systems
Sensors
- Vision-based inspection systems
- Monocular depth-estimation systems
- Autonomous positioning sensors
Protocols
- MAVLink
- Embedded serial communication
Results / Outcomes
- Successfully trained Mask R-CNN segmentation models for cable detection
- Implemented monocular metric depth estimation using Depth Anything V2
- Demonstrated autonomous vision-guided drone positioning
- Demonstrated computer-vision-based drone interaction using AprilTags
- Developed proof-of-concept autonomous infrastructure-inspection platform
Gallery
Engineering Notes
Autonomous Drone Platform
The inspection platform used a PX4-based autonomous drone system equipped with:
- onboard vision systems
- autonomous positioning capability
- infrastructure-inspection sensing
- computer-vision-guided interaction
The drone was designed for controlled positioning near communication-cable infrastructure.
Cable Detection and Cutting-Point Segmentation
A Mask R-CNN segmentation pipeline was trained to:
- detect cable bundles
- segment lashing-wire regions
- identify cutting points
- support autonomous interaction workflows
The segmentation system formed part of the autonomous cable-recovery perception pipeline.
Monocular Metric Depth Estimation
Depth Anything V2 was integrated to provide monocular metric depth estimation using visual imagery.
The system enabled:
- scene-depth understanding
- relative positioning
- infrastructure-distance estimation
- perception-guided navigation assistance
without requiring dedicated LiDAR hardware.
Vision-Guided Drone Demonstrations
AprilTag-based demonstrations were developed to validate:
- computer-vision-guided positioning
- autonomous approach behavior
- controlled interaction
- perception-guided drone motion
The system demonstrated stable autonomous approach and contact behavior using visual positioning techniques.
Related projects
More engineering case studies in this domain
Autonomous Pipe Inspection FPV Drone System
Developed an autonomous confined-space FPV drone system for pipe inspection using PX4 firmware modification, optical-flow fusion, laser-based centering, and YOLO-based defect segmentation.
AJNA Autonomous ICU Assistance and Telepresence Robot
Developed an autonomous healthcare robotics platform during the COVID-19 pandemic for ICU assistance, telepresence communication, patient monitoring, and autonomous medicine delivery.
Autonomous Solar Panel Cleaning Robot
Developed an autonomous robotic solar-panel cleaning system using computer vision, embedded control, and environmental sensing for industrial solar-farm deployment.