The Intelligent Systems subteam is responsible for automatically detecting and identifying targets from images captured by our plane in flight on our backend platform and, develop and test algorithms for mapping, warning systems and obstacle avoidance. For target identification, we collect images from our onboard camera and apply machine learning to accurately detect, localize, and classify targets by color, shape and alphanumeric. We use data visualization to test and view results. Our goal is to optimize system performance, and iterate and experiment with algorithms to better our accuracy and efficiency for mission objectives.

Ardupilot Parameter Estimation

Given a flight platform, we hope to output an optimal set of parameters used in the flight software we use (Ardupilot). The goal is to minimize the amount of time we spend tuning parameters with physical test flights.

Christian
Daniel
Sunny

Tracking

Given a set of geolocations, we want to change the position of the camera on board to consistently point at these spots on the ground. The algorithm outputs different values for roll and pitch of the gimbal as a function of variables like altitude, direction of plane, and distance to closest geolocation in the set.

Eric H
Anika