July 9 - Vanny for CV Auto Landings

Make sure to update the Title following the format “Jan 01, 2020 Flight Test”

 Flight Preparation

Requested by:

Megan Spee

Required Sub-teams

Mech
Electrical
EFS
CV
Operation
Sys-Int

Requested Date

Jul 3, 2023

Location and Time

July 9th 2023 @ WRESTRC 12 - 6 pm

Summary & Goals

  • CV Auto

    • Successfully detect the landing pad using Vanny

Approved?

Yes

  • Need camera, jetson, jetson power support

  • Need the jetson zip-tied on

  • Need jetson connections fixed up

  • have EE/CV assess the current wiring setup - may need to add a buck

 Required Persons During Test

Name

Phone #

Role

Reason

Name

Phone #

Role

Reason

 Megan Spee

 

Flight Test Lead

  • Creating documentation

  • Booking the test flight

  • Creating a thread in #flight-staging

 Anthony Luo

 

 Pilot in Command

  • Aircraft preparation

  • Filling out documentation

@Hardy Yu

 

Safety

  • Making sure everyone is safe

  • Bringing Med kit

@Amy Hu

 

Autonomy Technician

  • Check through the jetson system

  • Give flight test instruction

  • Records down the needed data

 Other Attendees

Name

Role

Name

Role

@Nathan Green

 

@Conall Kingshott

 

@Saransh Duggal

 

Drivers & Seat Arrangement

Driver

Vehicle Type

Passenger

Driver

Vehicle Type

Passenger

Anthony Luo

Rav4

 

Megan Spee

5-seater Prius

 

 

 

 

 

 Success criteria

Tests

Success Criteria

Requirements

Outcome

Tests

Success Criteria

Requirements

Outcome

  • Landing-pad model on Jetson

  • Landing-pad detection model runs on Jetson startup and saves the logging info with bounding box images in SD card.

  • computer-vision-python working and can run on startup

  • Jetson wired and mounted on Vanny correctly

Testing Timeline

Date/Time

Action

Date/Time

Action

July 7

Aircraft Assembly

  • Jetson system preparation

  • @Amy Hu Need to verify the CV system is working by Friday middle of the day

  • If Jetson can’t be tested and verified as working by Friday, the flight test won’t happen, and we will put the flight test to next week

July 8

Pre-flight Preparation

July 9, 10 am

Required persons for preflight preparation show up at bay

Flight test lead breifly assign task for preparation work

July 9, 11 am

Flight brief from flight test lead

July 9, 11: 15am

Lunch break

July 9, 11: 45am

 

Loading onto the Car

Take your water bottle with you. Go washroom before we leave!

July 9, 12:00

Drive to the test ground

July 9 12 - 3 pm

Flight Testing

July 9 3:15 pm

Debrief

Aircraft Incident Procedures

Incident

Procedure

Incident

Procedure

Crash

Outline the procedure for who will go to handle the aircraft in the event of a crash, what should be done, were emergency equipment is, etc.

Flyaway

Outline the procedure for flyaways. List relevant contact information for local authorities and plans of action to recover control of the aircraft

Configuration Error

Outline steps to take in the event of a configuration issue with the aircraft

Injury

 

Property Damage

 

Week-Before To-Do List

Drone Registration

Drone Registration Number:

Flight test location booked
Pilot scheduled
Test Plan (What we are testing)
This should also include the flight path where applicable
Subteam action items (for flight readiness)
Transportation logistics finalized
Booking SDC vehicles
Assigning drivers
Renting any needed equipment (trailer? generator? etc.)

Day-Before To-Do List

SDC Vehicle key collected (if required)
Batteries Planned for (DON'T LEAVE CHARGING OVERNIGHT)
On-site kit together
Airfame assembled
ESCs/flight controller wired
Radio equipment tested
Motors tested
Wiring secured (Harnessing)
End-to-end test (neglecting props)
Check the center of gravity
All components not on the airframe that are required should be in the flight box (e.g. props & prop nuts)
Airframe payload
Spare parts
Flight Code Flashed
Ground station Software updated

Pre-Departure Checklist

Pre-flight Safety Checklist

Post-Flight Checklist

Item

Owner

Item

Owner

ALL MEMBERS PRESENT

 

 

Flight Test Debrief

Flight test rundown:

  • Motor test

  • Flew in loit, failsafed due to GCS disconnect (rfd900’s)

  • Turned off GCS failsafe in pilot control mode

  • Flew over, landed on, over-ground landing pads.

  • Moved landing pads

  • flew over, landed on, over-grass landing pads.

  • Tried auto-tune on battery 1.

    • atune completed

    • thrust loss on 3

    • gps / compass glitch

    • “landing” on battery failsafe (3v/cell)

  • Tried auto-tune on battery 2

    • atune completed

    • landed & saved

Logs: 2032-07-09

Flight Test Post Mortem

  • 5W vtx delivers clear video.

  • Landing pad detection model works pretty well.

  • Autotune on YAW completed successfully twice.

  • Got lots of messages about Opflow/Lidar health. GPS1 also failing configuration checks

  • Batteries had a LOT of sag (21V in flight recovering to 23V after)

RCA for Red Items

Create a RCA and keep it as a child page.

5W VTX Performance

  • Noticed significantly better video transmission when landing in grass field south/south-east of skid pad.

    • No degredation in signal quality observed at all

      • comparable to fading to b&w and/or losing signal on 800mW.

Model Performance

 

  •  

     

Autotune Performance

  • Auto-tune completed successfully twice

  • The results were slightly different both times, maybe due to voltage difference between the two batteries:

 

 

Overall, we learned a lot about how autotune works, and are pleasantly surprised to see it function effectively. The general process is as follows:

  • Takeoff, fly over a certain area in a guided flight mode such as AltHold, Loiter.

  • Invoke autotune flight mode from the RC Controller

  • Wait for autotune to complete. Once done, the drone will use original tunes and maintain flight mode that it was invoked from.

  • Switch out of autotune into preferred flight mode for testing the tune.

    • If the tune is good → Switch back into autotune, land, and disarm. The values will save.

    • If the tune is not good → Land immediately and disarm, the values will not save.

Some important notes that we learned:

  • “Autotune: failing to level” messages can be safely ignored. Drift is not super bad with GPS hold.

Sensor Health

Throughout the flight test, there were numerous messages about bad opflow, or lidar health. Our suspicion is that these warnings are caused by the tall grass and trees creating variable and moving terrain under the drone.

Battery Sag / Voltage

Batteries sagged as low as 12V overall (2v/cell) near the end of the flight. The drop-off was incredibly fast as batteries were sitting around 21V/cell when autotune ended.

The low battery voltage led to a “supposed” lost of thrust on motor 3, where it was at full throttle and unable to provide thrust to change the attitude of the drone. This rapid lost in thrust would also contribute to our harsher-than-normal landing.

GCS Failsafe

Action Items

Other Resources

Booking WRESTRC Field

WRESTRC