Joint 69th Annual Southeastern / 55th Annual Northeastern Section Meeting - 2020

Paper No. 72-6
Presentation Time: 1:30 PM-5:30 PM

DIGITAL 3D MODELING OF ROCK SLOPES AT DIFFICULT SITES USING STICK-MOUNTED DRONES FOR IMAGE COLLECTION


EDGELL, Rebecca, RIMANELLI, Loukas, WITT, Hunter, STEPHENSON, George and WATTS, Chester F., Department of Geology, Radford University, Box - 6939, Radford, VA 24142-6939

The Radford University Geohazards Research Center is frequently tasked with 3D mapping of rock slopes along busy highways as part of rock slope stability investigations. Overlapping geotagged images are captured using drones and processed within structure-from-motion software to generate digital point clouds and triangle mesh models, from which geologic structure data can be extracted and plotted on stereonets. Drones are usually flown adjacent to the rock slopes, either manually or autonomously, using readily available mission planning software to assist in obtaining suitable imagery. At difficult sites along highways with narrow rights of way, overhanging branches, and heavy traffic, flying drones to effectively obtain slope toe imagery can be difficult, dangerous, a distraction to motorists, and possibly violate FAA regulations. A solution was proposed in which flights at traffic level were conducted by mounting the drone in a 3D-printed cradle attached to a pole or selfie stick and walked, rather than flown, along the shoulder on the far side of the rock slope of interest. Using the same drone as flown to image the upper portions of the rock slope ensured consistency of imagery, including resolution, exposure, and accuracy of georeferencing. The drone was programmed to take pictures every 3 or 6 feet of GPS-detected horizontal movement and every 6 or 9 feet of GPS-detected vertical movement. Test flights were conducted at a city park site in Radford, Virginia as wells as along a busy highway at Harpers Ferry in West Virginia. Slope toe-level images were combined with imagery from actual high altitude flights resulting in quality 3D point clouds and triangle mesh models, better suited for complete rock mass characterization than 3D models having missing or incomplete data from the base of the slope.