333A System for Safe Semi-Autonomous Indoor Flight

Опубликовано 21 июня 2016, 20:29

In this project, we present an end-to-end system for semi-autonomous flight of a Micro Unmanned Aerial Vehicle (UAV) in indoor, GPS-denied, environments. In particular, our system monitors a human user controlled flight and provides override commands to avoid collision with obstacles. We build on an established model for quadrotor dynamics to implement a system of nested control loops for attitude and position control. State estimates are calculated using an Extended Kalman Filter (EKF) to fuse onboard IMU acceleration and velocity measurements with Microsoft Kinect based RGBD visual odometry. A potential field based method that leverages point cloud data over a small time window is then used for obstacle avoidance. We demonstrate the effectiveness of our system in simulation and on an Ascending Technologies Firefly testbed