Crafting a Vision-Aided Software Stack for UAV
summary
A vision‑aided UAV mission server implements camera and gimbal hardware abstraction, RTSP video streaming via GStreamer, MAVLink integration for image capture and control, and a redundant data‑link using a MAVLink router combining 5G LTE and sub‑1 GHz telemetry. Prototype runs on Qualcomm RB5, uses hardware video/AI acceleration, interoperates with PX4/ArduPilot, and exposes camera/gimbal APIs; proof‑of‑concept works for streaming and command routing with plans to release the open‑source stack.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 2:32UAV mission server architecture
- 3:59Typical multi-rotor UAV system
- 5:47Embedded Linux on UAVs
- 7:54Software for deterministic missions
- 9:18Smart UAVs and task integration
- 10:24MAVLink protocol role
- 12:28Middleware for high-level tasks
- 13:13SLAM for UAV navigation
- 14:41AI and deep learning in UAVs
- 15:45Open-source autopilot systems
- 17:13Autopilot state estimator
- 18:40ROS vs LCM
- 21:33Camera and gimbal challenges
- 23:21Redundant MAVLink router
- 24:03Mission server proof-of-concept
- 25:51Multi-core flight control future
- 26:33Q&A
related talks
18mFlying drones with Twitter, Bananas and WebAPIs
40mApplication and Challenges of Using GenAVB/TSN Software Stack for Industrial...
33mFlying a drone with twitter, bananas & Web APIs
37mDevelop modern connected applications on 5G MEC/Edge Space and Azure
29mFlying a drone with twitter, bananas & Web APIs
43mTo kill a working drone: flying a drone with twitter, bananas & Web APIs
28mDrone economy 2.0: Economic use by digital visualization.
38mFlying a drone with twitter, bananas & Web APIs