Augmenting Inspection Capabilities with Mixed-Reality
We demonstrate the use of mixed-reality headset worn by an inspector on the site of inspection. Such a headset along with the software we have developed, can augment the on-site inspector's capabilities to visualize and interact with digital annotations. Thus allowing precise and clear information about the region of interest.
Demo Run: Thermal, Monocular Vision-Based Localization and Mapping of MAV at Night.
Video runs at 0.5x. This video demonstrates how the state-of-the-art monocular computer vision algorithms perform in thermal IR domain.
S. Daftry, M. P. Das, J. Delaune, C. Sorice, R. Hewitt, S. Reddy, D. Lytle, E. Gu, L. Matthies. Robust Vision-based Autonomous Navigation, Mapping and Landing for MAVs at Night. International Symposium on Experimental Robotics (ISER), Buenos Aires, 2018.
5-DoF Monocular Visual Localization Over Grid Based Floor
Quadcopter mapping, path planning and controls for Obstacle avoidance.
Sensors: 2D Lidar, GPS, IMU
Flight controller: PX4
Path planning algorithm: RRT# using OMPL C++ (http://ompl.kavrakilab.org/)
Other libraries used: FCL (A Flexible Collision Library), Octomap, ROS.
Simulation is made in Gazebo.
VINS-Mono Visual Inertial Odometry on Optor Stereo Camera