Admin Preparation

Testplan

This FT will include two separate Flight:

• Frist one: trying to tune Houston

• Second one: Perform autonomy Tasks

Note: The Second flight test will continue if only if the first FT is successful.

Necessary Preparation

Necessary items:

• RPI

• Lidar

Flight characteristics needed:

• Flight stability

Comms / Support needed:

Mandatory Attendees

Flightline Team

Pre-Flight Checklists

Flight Test Timeline

DEBRIEF

Flight test debrief report

Flight Test #1 - Autotune (Morning):

• Flight #1 (Performance Check) - Takeoff 10:32

  • @Yuchen Lin took off, checked performance of drone

  • Noticed issue where drone was drifting and also yaw was twitching when turning/inputting throttle

  • This is ok, as goal was to tune the drone.

• Flight #2 (Quicktune) - Takeoff 10:39

  • @Smile Khatri took off, engaged the quicktune script

  • The script did not engage - @Yuchen Lin to investigate root cause

  • Bit of confusion regarding if it was quick tuning or not as Smile was inputting and Yuchen thought this was quicktune working

    • Should be more clear communication between GSO and pilot

    • When tuning inputs should be kept to a minimum

  • Onsite debugging… LUA code is updated but issue is the engage message does not show

  • Decision made to just complete a full autotune

BATTERY SWAP

• Flight #3 (Autotone for Roll and Pitch) - Takeoff 10:52

  • @Evan Janakievski took off and engaged autotune

  • Roll tuning occured first, progressed normally, no issues here

  • Then pitch tuning occured

    • Was almost finished but then did not give a confirmation message

    • @Yuchen Lin made decision to just land and take current setting

    • Once landed we saw finished message, so issue resolved

  • Landed in autotune to save PID settings

BATTERY SWAP

• Flight #4 (Autotune for Yaw D & E) - Takeoff 10:58

  • @Manasva Katyal took off and engaged autotone

  • Tuning occurred properly, no issues observed

  • Fixed the yaw twitching issue observed in Flight #1

BATTERY SWAP

• Flight #5 (Pilot Training) - Takeoff 11:12

  • @Manasva Katyal @Smile Khatri @Evan Janakievski pilot training

  • All observed better response from drone, stable in loiter

  • No yaw twitching issues

  • Overall good, “It was smooth”

• RCA for Last Week:

  • Last week’s autotune issues stemmed from bad notch filter

  • @Yuchen Lin tuned and corrected this

Flight Test #2 - Auto Obstacle Avoidance & Geolocation (Afternoon):

• Preflight Testing Note:

  • Was a significant amount of delay in getting out of the bay

  • We should have been out by 1:30

  • Ended up leaving at 3 PM

  • Suggestion to ground test more complex things like autonomy’s requests today in advance of flight tests such that this does not occur

• Flight #1 (Validation Attempt): - Takeoff 15:26

  • @Evan Janakievski took off in Loiter

  • Only input was throttle > 50% and it started drifiting

  • Yawed wildly

  • Immidiate termination by bringing drone down (no need for disarm in flight)

  • Everytime we mount autonomy stuff we see this happening

    • Don’t know if it is LIDAR or RPI

  • EKF issue, performed recalibration

  • Took a while because we were doing it near vehicle and cellphones

  • Once we moved to open, clear area proceeded smoothly

  • Also made MAG 1 enabled and MAG 2 disabled (before were using MAG 2)

  • This resolved issues as seen in next flights

  • What we suspect:

    • LIDAR or RPI causing magnetic deviation for compass

    • This caused the fishbowling observed in this flight

    • So recommendation is to recalibrate IMU & Compass after any major mounting changes, especially on a smaller drone like Houston with close proximity mounting

• Flight #2 (Validation Attempt 2) - Takeoff 15:52

  • @Yuchen Lin took off in Loiter

  • Did not observe any compass variance this time

  • Overall pretty successful in terms of validation

  • At landing throttle was cut a little too much a little too early

  • This caused a loss of thrust on landing

  • To confirm this was pilot error and not a drone error, recommended another small validation flight

• Flight #3 (Throttle Validation) - Takeoff 15:55

  • @Yuchen Lin took off in Loiter

  • Verified throttle performance

  • Was good, made decision to proceed with Auto testing

BATTERY SWAP

• Flight #4 (Auto Obstacle Avoidance Dry Run, No Obstacle) - Takeoff 16:02

  • @Manasva Katyal took off in Loiter

  • Engaged Auto to check if waypoints worked corrected

  • Everything proceeded as expected flightwise, nothing much to say here

  • However the 12V to 5V buck converter broke (USB-C connector)

    • It was clearly not soldered properly… @Daniel Puratich @Meghan Dang RCA

• Flight #5 (Auto Obstacle Avoidance Test) - Takeoff 16:08

  • @Manasva Katyal took off in Auto, but program forced Loiter as LIDAR detected something (probably the ground)

  • When we were ground testing, we raised drone up and it went into Auto so we know this works

  • However as we rose higher and higher the drone never switched into Auto

  • This suggests LIDAR is still detecting something

  • @Andrew Shum to RCA

  • Did a lot of ground testing for this but could not get anywhere

    • Narrowed the LIDAR spin angle from 180 to 90

    • Changed distance to 4m

    • Wiped the LIDAR with chamois cloth

    • Also an important note is phone hotspot SSH was quite unstable… need to look into better solutions @Andrew Shum @Yuchen Lin

  • Made decision to postpone this and proceed with Geolocation testing

• Flight #6 (Auto Gelocation Data Collection @ 3m) - Takeoff 16:47

  • @Evan Janakievski pilot flying, took off in Loiter

  • Performed geolocation testing at a height of 3 meters

  • After about 2 minutes observed very high current draw due to high weight and hot conditions

    • Was at 26 amps

    • For reference in the morning flight test was at 17-19 amps

  • This called for termination but when landing at ~1 meter drone went into battery failsafe and tried to perform a RTL by entering that mode

  • Pilot reported loss of control, rectified by switching back to Loiter

    • Important point for pilot training

  • Also battery seemed a little puffy… will be disposed

    • On this note we need to look into getting new batteries @Yuchen Lin

  • Another issue was missing screws on one of the legs

    • @Smile Khatri @Evan Janakievski determined root cause, rectified

BATTERY SWAP

• Flight #7 (Auto Geolocation Data @ 7m & 10m) - Takeoff 16:57

  • @Smile Khatri took off in Loiter

  • Flew at 7 meters for 1.5 minutes

  • Then handover to @Manasva Katyal in the air

  • Flew for 2 more minutes

  • Kept it quick because of battery failsafe scare from prior flight

  • Note that on hot days with this level of current draw we have basically no flight time

• Flight #8 (Final Auto Obstacle Avoidance Test) - Takeoff 17:15

  • Final test with LIDAR detection distance at 2.5 meters

  • @Manasva Katyal took off in Guided mode and drone starting navigating properly

    • This is because arming in Auto was not allowed

  • However the script terminated due to leaving Auto/Loiter modes, as it was programmed to do

  • So unable to determine anything from this

  • Future attempts called off as it was getting late, will be pushed to next flight test

• Overall Recommendations:

  • Need to conduct more thorough ground testing for the obstacle avoidance

  • Also need to do some LIDAR testing/benchmarking

  • Two possible causes here:

    • LIDAR malfunctioning (aka “lidar not happy” - Yuchen)

    • Something is in the way and being detected (ex. props)

      • To ensure this is not the case, @Smile Khatri will look into elevating LIDAR more

  • Actionables:

    • Test with Pixhawk proximity sensor setup

    • See if LIDAR can get consistent results under 10 meters

    • Make a Plan A/B to get meaning for result out of future flight tests

    • Start with Waypoint 1