rf silent drone navigation
play

RF Silent Drone Navigation ANNA JAMES OUR TEAM Matthew Dupree - PowerPoint PPT Presentation

RF Silent Drone Navigation ANNA JAMES OUR TEAM Matthew Dupree Xihan Liu Yingchao Zhu Student Lead Data Analysis PCB/Schematic 01 OVERVIEW Liftoff from a boat at sea Land on a boat at sea Limited-size landing area Moving


  1. RF Silent Drone Navigation

  2. ANNA JAMES OUR TEAM Matthew Dupree Xihan Liu Yingchao Zhu Student Lead Data Analysis PCB/Schematic

  3. 01 OVERVIEW

  4. ❏ Liftoff from a boat at sea ❏ Land on a boat at sea ○ Limited-size landing area ○ Moving landing area Overview ○ Miss == wet hardware ❏ RF-silent: No radio communication ○ No pilot ○ No landing beacon ○ No external processing ➢ Computer vision is hard on an energy budget!

  5. Problem Formalization

  6. Camera: Electronic Power Management Raspberry Pi Camera Speed Motors 3-Phase Board Module V2 Controllers Parallel PWM Block Diagram Companion Computer: Flight Controller: UART Raspberry Pi 3B+ Pixhawk 2.4.8 Lipo Ethernet Battery I2C GPS: Micro M8N GPS Module LiPo to 5V Voltage Regulator

  7. Camera: Electronic Power Management Raspberry Pi Camera Speed Servo Motors 3-Phase Board Module V2 Controllers Parallel PWM Block Diagram Companion Computer: Flight Controller: UART Raspberry Pi 3B+ Pixhawk 2.4.8 Lipo Ethernet Battery I2C GPS: Micro M8N GPS Module LiPo to 5V Voltage Regulator

  8. 02 Hardware

  9. Three-layer design ❏ Top: GPS, Flight Controller, Telemetry radio ❏ Middle: Companion Computer, Camera ❏ Bottom: 4s Lipo Battery

  10. Hardware Pixhawk v2.4.8 ❏ Flight control unit (FCU) ❏ Uses ArduPilot firmware ❏ Ensures drone stability Top view in flight

  11. Hardware uBlox M8N Micro GPS ❏ GPS unit ❏ Provides drone lat/long coordinate reference Top view ❏ Used until landing target detected

  12. Hardware Transceiver Telemetry Radio ❏ Broadcasts debugging information to GCS ❏ Used for our testing -- not Top view required for flight

  13. ❏ Middle: Companion Computer, Camera

  14. Hardware Raspberry Pi 3B+ ❏ Companion computer ❏ Performs CV tasks to identify the target and Side view communicate target transform to FCU

  15. Hardware Raspberry Pi Camera Module v2 ❏ 8 megapixel camera capable of taking photographs of Bottom view 3280 x 2464 pixels ❏ Used to detect AprilTags in 10FPS 640x480 mode

  16. ❏ Bottom: 4s Lipo Battery

  17. 03 Printed Circuit board (PCB)

  18. Schematic ❏ Function as a voltage regulator to convert the input voltage from 14-22V to 5V output voltage ❏ Diode after the input voltage in order to prevent reverse current

  19. PCB board

  20. ❏ 2-layer PCB PCB Board ❏ 45 * 59 mm ❏ Holes for thermal dissipation

  21. 04 Software

  22. Landing Target AprilTag 3 ❏ Landing Detection Target ❏ Works on resource-constrained platforms (like our RPi 3B+) ❏ Full transforms from single stills!

  23. Stats With current onboard Tag Images processing, we have tested that at 5 meter altitude the craft can search at 10FPS in a 5x6.7 meter box below the vehicle for a 16.5cm square (Example photo taken at ~2m tag. alt. and has been cropped.) Could do far better with more processing power!

  24. First-choice: Old but gold: ArduPilot Multicopter FCU Firmware PX4 Autopilot ❏ Great droning OS ❏ Heavily used ❏ Great debug tools ❏ Many configurations ❏ Advanced features (eg MAVLink Shell) ❏ Spaghetti-code ❏ Doesn’t support our ❏ Difficult configuration precision landing use! ❏ Missing debugging shells

  25. Ubiquity Robotics’ Ubuntu Mate 18.04 Companion OS With ROS Kinetic ❏ All open-source parts ❏ Free to use ❏ Spreads tasks among processes to make best use of hardware ❏ Lots of logging tools!

  26. Legend Raspberry Pi Hardware Camera Module ROS Node Processing Pipeline RasPiCam_node Data Images Debugging ❏ A large stack of software Software Apriltag_ROS packages from the Robot Tag Positions Logging Operating System (ROS) Vision_to_MAVROS Pipeline V_sub.py Target ecosystem (Our custom Logging Positions visualizer) ❏ High-throughput MAVROS PixHawk COTS communication between Logging FCU nodes via ROS topics MAVExplorer ❏ Relies on many complex Packages in bold we configuration files had to largely rewrite or write entirely ourselves!

  27. Companion Flight ROS/MAVLink/UART Computer Controller Location Data ❏ Drone location found MAVLink Shell relative to the camera YAML ROS topic Data EXTRACTION Converter SSH Pipe echo position Pipe Shell ❏ Data points are converted to angle and v_sub.py distance and North-East- Down Position Rotation ❏ For debugging, our v_sub.py plots the streaming data onto a Matplotlib 3D plot

  28. 05 Demo

  29. Special Thanks to: Acknowledgements Navsea, Project Sponsor Alan Jaeger, Navsea Representative Dr. Yogananda Isukapalli, CE Capstone Project Instructor Aditya Wadaskar, TA Kyle Douglas, TA

  30. THANKS!

  31. Questions?

Recommend


More recommend