Vision-Based Collision free technique for Autonomous Navigation of MAV in GPS-denied Environments
2022-26-1250
05/26/2022
- Event
- Content
- The drone-based aerial platform offers various applications such as security and emergency response, delivery, defense, inspection, mapping, film, and entertainment. For safe driving, an automated vehicular navigation system (AVNS) must have a precise location. Although the vehicle location can be retrieved in GPS-friendly surroundings, the positioning accuracy of AVNS is easily reduced in GPS-denied environments (such as suburbs, tunnels, forests, and indoor scenarios). Researchers have already developed various methods to handle UAV navigation in GPS-less environments, such as Visual Odometry (VO) and Simultaneous Localization and Mapping (SLAM). This paper proposes a new Qualcomm Snapdragon referenced visual-inertial odometry (VIO) algorithm for indoor GPS-denied Micro Aerial Vehicle (MAV) navigation. The custom-built MAV is comprised of a hardware platform equipped with a VOXL flight companion computer, PX4 flight controller, pre-calibrated stereo and tracking sensors, a high-resolution first-person view (FPV) image sensor, and spectrum RC telemetry receiver. The data from the stereo sensors are utilized to calculate the depth of mapping, which aids in collision avoidance. The tracking sensor data can develop visual-inertial odometry (VIO) algorithm for indoor navigation. The indoor flight experiment demonstrates that the MAV with the proposed algorithm can navigate fully autonomous missions without collision while establishing accurate surrounding maps.
- Citation
- Duba, P., "Vision-Based Collision free technique for Autonomous Navigation of MAV in GPS-denied Environments," SAE Technical Paper 2022-26-1250, 2022, .