BattleDrones
Custom-built drones with AI capabilities to be tested at the first-ever CCI BattleDrones competition
Decision-making and computer vision with machine learning can be used to produce spacecraft controllers and decision matrices for deep space missions. Path planning optimization can be used for both robotic bodies and robotic arms in and around structure or defined space, using both GPS and onboard sensors.
Rendered video of competition gate using Blender
Automation
Using a YOLO3 neural network trained on procedurally-generated videos, the drone learns to recognize obstacles. I focused on developing the autonomous navigation and control algorithms necessary to evade obstacles and future offensive strategies.
Virginia Tech’s drone cage, where the obstacle course will be built
BattleDrones
In spring 2022, Virginia Tech will host Commonwealth Cyber Initiative’s first BattleDrones competition. Teams are provided with a custom-build base drone architecture that can be augmented to increase performance.
On-board NVIDIA Jetson Nano and camera
Decision-Making
Future iterations will include ML Markov Decision Process algorithms to optimize path planning and avoid cyber-physical obstacles.