Exploration and Object Detection via Low-Cost Autonomous Drone
Branden Pinney, Ben Stockett, Mohammad Shekaramiz, Mohammad A. S. Masoum, Abdennour Seibi, Ángel Gaspar González Rodríguez
Abstract
This work presents an initial study on developing an autonomous drone path-planning solution to locate and inspect wind turbines in a wind farm. Here, we show the results of an offline exploration solution to identify and approach targets with unknown locations in the area of interest. This will be useful for missions with a long-range traverse of a drone, in which as the drone is reaching the vicinity of the wind turbine, real-time GPS information may not necessarily lead the drone to reach right in front of the desired turbine. In this paper, our investigation aims at comparing the effectiveness of two area exploration algorithms through their battery usage, time to complete the pattern, distance traversed, and the time required to detect several targets placed in the area of exploration. The targets in this study are metallic pedestal fans acting as a surrogate for full-size wind turbines. A low-cost drone was used for small-scale laboratory experiments indoors. The targets are identified via object detection and an attached QR code at the base of the pedestal fans for target verification.