Trajectory optimization methods for drone cameras

Digital content

Author/Creator: Roberts, Michael Lindsay, author.
Contributor: Hanrahan, P. M. (Patrick Matthew) degree supervisor.; Durand, Frédo, degree committee member.; James, Doug L. degree committee member.; Stanford University. Computer Science Department.

Summary: Drone cameras are now being deployed in a wide range of applications, including Hollywood filmmaking, search and rescue, wildlife monitoring, and large-scale 3D scanning. However, drones remain difficult to control, both for humans and for computers. In this dissertation, we introduce a variety of trajectory optimization methods that make it easier for people to use drone cameras. We focus specifically on two different applications: drone cinematography and drone 3D scanning. Throughout this dissertation, we leverage domain knowledge that is specialized to each application, and we reformulate classical trajectory optimization problems in terms of the drone camera's visual output, i.e., in terms of what the drone is seeing.