Research on an Autonomous Vision-Guided Helicopter

We present an overview of the autonomous helicopter project at Carnegie Mellon’s Robotics Institute. The goal of this project is to autonomously fly helicopters using computer vision closely integrated with other on-board sensors. We discuss a concrete example mission designed to demonstrate the viability of vision-based helicopter flight and specify the components necessary to accomplish this mission. Major components include customized vision processing hardware designed for high bandwidth and low latency processing and 6-degree-of-freedom test stand designed for realistic and safe indoor experiments using model helicopters. We describe our progress in accomplishing an indoor mission and show experimental results of estimating helicopter state with computer vision during actual flight experiments.