Software Paves the Road Forward

MAAV’s computer vision division within their software subteam met on October 25th. The team is in the early stages of the year, assigning different projects to newer members as they complete their software tutorials. Cheng, one of the leads, walks through the various computer vision subprojects that need to be completed to complete the mission for the international aerial robotics competition (IARC). These subdivisions are each implementing a different software feature for the vehicle — camera calibration, line detection, roomba detection and RGBD sensors. Most are involved in augmenting our drone’s vision and decision-making.

The current IARC mission involves our vehicle interacting with randomly moving ground robots (roombas), a difficult task that previous missions did not feature. To ensure a solid performance by our vehicle, the software team is developing algorithms to support various forms of detection to aide the decision-making of our A.I. A significant change they are implementing is the upgrade from RGB sensors equipped on our vehicle to RGBD to improve its decision-making algorithms by providing better information on image depth.