SURG Fall 2011¶
Aerial Point-Cloud Generation using the Microsoft Kinect on an Autonomous Quadrotor¶
Project Members¶
Jeff Cooper, Jitu Das, Priya Deo, Daniel Jacobs,
Mike Ornstein, Harrison Rose, Alex Zirbel.
Advisor¶
Sanjiv Singh
Abstract¶
A robot is most useful when it not only understands and responds to its environment, but is
able to convey its information to a human operator in a useful way. This research aims to
tackle the simultaneous goals of navigation in a tightly constrained indoor area and the
transfer of useful information in a format designed for human use. The quadrotor aerial
platform provides an attractive solution to the navigation problem: being extremely
maneuverable, powerful enough to carry high-performance sensing equipment, and compact
enough to fit through small gaps such as a broken window, a quadrotor can gather data about
a location before humans or other robots enter it. Microsoft’s Kinect sensor provides both
color and depth data at an extremely low cost, which the robot can gather and transfer to an
off-board unit for processing. Once this data is available, point-clouds can be used to
display each reading and corresponding color in a 3D map which a human operator could
visually understand. The combination of autonomous exploration, 3D data collection, and
useful display is the principal goal which this project will attempt to achieve.