Michal Podhradsky
Dissertation
A Multi-Agent System for Adaptive Control of a Flapping-Wing Micro Air Vehicle
This page provides supplementary materials for my dissertation. Below you will find annotated videos from the experiments, as well as my source code. Should you have any questions, don't hesitate to contact me.
Download
You can download my dissertation here [PDF]
Source Code
Source code is available here
Latex Template
Latex template for my Portland State University dissertation is available here
Relevant Publications
- An Evolutionary Approach to Tuning a Multi-Agent System for Autonomous Adaptive Control of a Flapping-Wing Micro Air Vehicle, Michal Podhradsky, Garrison Greenwood, John Gallagher, International Journal of Robotics and Automation Technology (2016) [PDF]
- A Multi-Agent System for Autonomous Adaptive Control of a Flapping-Wing Micro Air Vehicle, Michal Podhradsky, Garrison Greenwood, John Gallagher, Eric Matson, IEEE International Conference on Evolvable Systems, (ICES 2015) [PDF]
Video
Dissertation defence
Recorded live stream from the dissertation defence.
Introduction
An introductory video showing the robot, the experimental setup and a running experiment.
Waypoint following
The vehicle is placed in the middle of the water tank, and after the initialization it goes back and forth between two waypoints (marked with green dots). See Section 5.3 of my dissertation for more details.
Fault detection and recovery
The vehicle is detecting and recovering from the Left Wing damaged fault. We cut down the left wing by 30 percent. After initialization, the vehicle rotates right by 360 degrees and measures time needed to perform the maneuver. Based on the duration of the maneuver the likelihood of a fault is evaluated. If a fault is detected (which is this case), updated recovery values of deltas and omegas are loaded from the memory. Then the vehicle continues going to the destination waypoint. See Section 5.4 of my dissertation for more details.
Obstacle avoidance
A virtual obstacle (marked as a green line) is placed in the middle of the water tank. The vehicle is commanded to go to the waypoint at the opposite side of the water tank. Because the waypoint is behind the obstacle, an obstacle avoidance routine (described in Algorithm 3) is performed. See Section 5.5 of my dissertation for more details.
Example of remote control
An example of the early version of GUI is shown. In this video we tested remote control of the vehicle - by moving the sliders, the wings respond accordingly. See Appendix B for more details.