Related Work
Related Research
There have been numerous uses of blimps for robotics and controls classes, including here at the UW. In Spring 2006 Dieter Fox taught the CSE 481 Robotics Capstone using a blimp project. This project seems a bit larger in scope than ours, having spent the entire quarter on the blimp. An overview of their outline shows that they applied many topics we covered briefly in class: probabilistic sensor models, probabilistic state estimation, Gaussian filtering, Kalman filter, and PID Controls. They also went into some deeper topics such as path planning, Simultaneous Localization and Mapping (SLAM), and Markov Decision Processes (MDPs). The course website can be found here: http://www.cs.washington.edu/education/courses/cse481c/06sp/
In addition to blimps, there is similar related work with quadrocopters. These are miniature UAV “drones” with four motors and propellers. These vehicles maintain elevation with prop-thrust rather than buoyancy, yet they present many of the same interesting challenges as our project: radio communication, proximity detection with set-points, PID controls, etc. The quadrocopter platform is even more difficult than the blimp since it is inherently unstable and falls out of the sky when not working properly. Multiple accelerometer and MEMS gyros are sensed with PID control loops to maintain proper vehicle attitude and enable flight.
In Spring 2011 Swetak Patel taught the CSE 477 Embedded Systems capstone using the Parrot A.R. Drone, a commercial hobbyist RC quadrocopter. Dr. Patel also uses this platform for the EE 472 Embedded Systems course that he teaches. In these courses students explore a variety of modifications to the drone system, including adding addition proximity sensors (sonar), adding machine vision techniques, wireless power transfer, and developing their own control-input systems. The course websites can be found here: http://abstract.cs.washington.edu/~shwetak/classes/cse477/ http://abstract.cs.washington.edu/~shwetak/classes/ee472/