Detection of Long Narrow Landing Features for Autonomous UAV Perching

It is highly advantageous for Unmanned Aerial Vehicles (UAVs) to be able to perch on natural features or human-made structures in order to conserve power, maintain a fixed position for data collection, or for performing manipulator tasks. Long narrow landing features can commonly be found on industr...

Full description

Saved in:
Bibliographic Details
Main Authors: von Frankenberg, Florentin, Nokleby, Scott
Format: Conference Proceeding
Language:English
Subjects:
Autonomous
Manipulators
Mobile communication
Perching
Poles and towers
Power transmission
Task analysis
Three-dimensional displays
UAV
Unmanned aerial vehicles
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:It is highly advantageous for Unmanned Aerial Vehicles (UAVs) to be able to perch on natural features or human-made structures in order to conserve power, maintain a fixed position for data collection, or for performing manipulator tasks. Long narrow landing features can commonly be found on industrial structures such as power transmission towers and are convenient for UAVs to perch on due to being easy to grasp and being generally unobstructed by the presence of other nearby objects. A landing feature detection algorithm was developed and implemented using a Microsoft Kinect to provide 3D point cloud data. The algorithm was tested on a scene composed of two step-ladders and several rods joining the two together. The algorithm detects line segments in the point cloud data using Random Sample Consensus (RANSAC), and then evaluates the suitability of the detected lines for landing. The algorithm successfully detected rods which were at a suitable orientation, the correct diameter and length, and which were unobstructed by other objects.
ISSN:2644-3163
DOI:10.1109/IEMCON51383.2020.9284856