[Thesis Defense] Tools to Facilitate Autonomous Quadrotor-Based Cinematography





The interactive transcript could not be loaded.



Rating is available when the video has been rented.
This feature is not available right now. Please try again later.
Published on Dec 16, 2016

I defended my PhD Thesis on December 12th, 2016 at Stanford University. It went really well, I had an overwhelming outpour of enthusiasm and excitement from the 50+ friends and colleagues that came to see my work. Thank you very much everyone!

My committee consisted of:

Prof. Mac Schwager (Chair), AeroAstro
Prof. Pat Hanrahan (PI), Computer Science
Prof. Maneesh Agrawala, Computer Science
Prof. Stu Card, Computer Science (Consulting Professor)
Prof. Juan Alonso, AeroAstro
Dr. David Merrill, Computer Science (External Member)

Title: Tools to Facilitate Quadrotor-Based Cinematography


There is considerable interest in using orientable cameras mounted on unmanned quadrotor aircraft for cinematography. Quadrotor cameras can fly to unique vantage points and execute dynamic camera moves in 3D space. However, traditional methods of manually piloting quadrotor cameras to capture video do not consider the needs of cinematographers, and require sophisticated skill and dexterity. In this thesis, we present an alternative approach which reifies concepts from traditional and virtual filmmaking into tools for cinematographers. These tools enable the user to express their cinematic intent directly, while we automate flying the quadrotor camera.

We begin by describing a tool for designing and autonomously executing quadrotor-based camera shots. Our tool enables the user to visually compose shots using keyframes, and precisely specify shot timing using easing curves. The user can preview her shot in a virtual environment before flying, and automatically capture shots in the real world with a single button click using commercially available quadrotors. Since some visual compositions of shots are more favorable than others, we next present a tool that automatically computes static shots based on well-established visual composition principles and canonical shots from the cinematography literature. Our tool calculates feasible, safe, and visually pleasing transitions between shots using a novel real-time trajectory planning algorithm. Using our tool, the user can capture shots that follow cinematic conventions without spending cognitive effort on setting up individual compositions, or specifying how to move between compositions. Finally, because a quadrotor camera must also know where subjects are with sufficient accuracy to faithfully capture these canonical shots of people, we describe a platform for accurately localizing multiple objects. By using RTK-GPS and IMU sensors, our platform provides centimeter-accurate tracking and decimeter-accurate quadrotor control in a large-scale outdoor environment. In combination, this work enables novices and experts alike to capture high quality video footage using quadrotors.


When autoplay is enabled, a suggested video will automatically play next.

Up next

to add this to Watch Later

Add to

Loading playlists...