The SkyNET drone created by students Theodore Reed and Joseph Geis under the guidance of Prof. Sven Dietrich as a proof of concept for a research project on side-channel botnet command and control. The research was presented at Usenix WOOT 2011. The project consisted mostly of software design to utilize a 3G/GPS-controlled UAV to enlist and command a hypothetical botnet. Although the drone does have malicious potential, it was not, and never will be, used for malicious purposes.
Most of the controls were programmed on the TS-7552 SBC, this board allowed us to add a 3G modem, GPS, and multiple WiFi cards to the AR.Drone. The added payload (cards, board, etc) is about 170g. The drone flies for about 13 minutes continuously, and for about 2.5 hours while grounded. The concept we demoed was a drone which would use half of its power to position itself, and the remaining to perform command and control.
The main benefit of using a modified retail toy was to attract potential victims, and remain innocuous during urban flight (not to mention the reduced cost). The drone can be replicated for under $500 US.
Random Motion and Collision Detection in 3D In this class project, event driven OpenGL programming is used to implement random motion in 3D. Collision detection ensures that all objects, both mobile and stationary, behave intuitively in a gaming environment. Work performed by Olga Koteoglou under the guidance of Prof. George Kamberov in Fall 2011 CS 537, Interactive Computer Graphics.
Hierarchical Scene Rendering As a demonstration of foundational graphics techniques, this scene consists of a hierarchy of layers, each layer consisting of simple shapes. Textures are added for realism. The user may navigate the scene from inside a car whose windshield is transparent glass. Work performed by Olga Koteoglou under the guidance of Prof. George Kamberov in Fall 2011 CS 537, Interactive Computer Graphics.
Reciprocal Collision Avoidance algorithm prevents collisions among 800 running agents as each runs to one of 4 goal positions. The agents must avoid each other as well as 9 static obstacles. Work performed by Yang Wang under the guidance of Prof. George Kamberov in Spring 2012 CS 586, Machine Learning for Game Design.
Portal Culling algorithm considers the position of the viewer and renders only portions of other rooms that can be seen through portals from the viewer's current position. Work performed by Lazaros Karydas under the guidance of Prof. George Kamberov in Spring 2012 CS 539, Real Time Rendering, Gaming, and Simulation Programming.
A collection of examples of how to create realistic natural environments from algorithms. Examples are: steering behavior of a flock of birds, reflection of surroundings onto a regular shape such as a sphere, and reflection of surroundings onto a simulated lake surface. Work performed by Lazaros Karydas under the guidance of Prof. George Kamberov in Spring 2012 CS 539, Real Time Rendering, Gaming, and Simulation Programming.
Procedural Modeling is a technique that generates many random instances of a single model; for example, to create many slightly different buildings to form a city. Crowd Simulation allows agents to move freely toward a destination while avoiding static obstacles and other moving agents. Work performed by Hao Kang and Cory Spencer under the guidance of Prof. George Kamberov in Spring 2012 CS 586, Machine Learning for Game Design