2023-24 Summer Auto-Landing Project
Goal
Construct an auto-landing algorithm to compute the distance vector from the drone to the landing pad. The distance vector will be converted into a movement command that will be sent to ZeroPilot (WARG’s in-house flight controller).
Diagram
Â
Â
Fisheye Distortion (compensation) this can be fixed CV camera is rectilinear lol
Â
Landing Pad team, if a landing pad is detected with 90% confidence, send bounding box info to auto landing code.
We need all code to be put into the UWARG Github
Get inputs
from ZeroPilot
Landing pad team will get bounding boxes for you guys
Task: Get people to work on calculating angle of drone to landing pad
You have (pitch, yaw, roll) of drone
Camera is mounted at the bottom of the drone and is parallel to the bottom (Talk to mech)
Angle as seen from the camera (comes from the bounding boxes)
Task: Integrate 1D LiDAR and get altitude of drone
Note: It will be pointing at the bottom and it will parallel to the bottom (Talk to mech)
Â
Things to consider
Account for turbulence of the drone
Account for momentum of the drone. At every point in time there is a velocity vector associated with the drone. So plan the distance vector accordingly.
Velocity could be estimated via past location of the drone
Action Items
@Michael Denissov discuss with EFS what they would need from Autolanding on the Autonomy side
@Amy Hu Verify the camera is rectilinear
@Mihir Gupta (Unlicensed) organize this confluence page and separate the project into tasks
@Michael Denissov Organize all the code from other people’s repo and @Mihir Gupta (Unlicensed) will create an autolanding repo on UWARG github