2024-12-14 Pegasus 2 Manual Piloting Flight Test
- 1 Planning
- 1.1 Scope
- 1.2 Flight Cards
- 1.3 Timeline
- 1.4 Attendees
- 1.4.1 Key People Not Attending
- 2 Debrief
Planning
Scope
Pegasus 2
Competition Task 2 Testing
Manual piloting, no autonomy
Test both solutions, both should be hot swapable and swapping between them validated in the bay in advance
Houston
Competition Task 1 - full system test (with pilot flying)
Tracking Antenna - Pushed to TBD, still issues with sending data.
Will test Pitch and Yaw
Flight Cards
Flight Card 1
Pegasus 2 Competition Solution 2
Pilot: @Nathan Green
GSO: @Evan Janakievski
What is being tested? To be filled out
Water Pump
Angled landing gear
Hardware to be installed, items to bring, batteries, etc
2 pumps and tank
8 6s batteries
Water controller boards
Drum
Drum fixture
TEST PROCEDURE (PEGASUS 2) | ||
Procedure | Goals / Objectives | Knockoff criteria |
| Test control and handling of Pegasus' tune with empty payload | Issue with solution mechanically or electrically |
| Prove new landing gear holds up with landing on drum without water risk | Landing gear breaks, can’t land on drum |
| Prepare for next step | Drum leaks? |
| Practice |
|
| Prove pumping works | Issue with water pumping, too much leaking, burned out pump |
| Test tune with extra weight | Flies bad |
| Test tank emptying works without having to hover | Issue with water pumping, too much leaking, burned out pump |
| Prepare for next test | Same as above |
| Test pumping out while hovering | Same as above, jittery, battery problems |
Repeat as needed or as batteries allow |
|
|
Flight Card 2 not ready - pushed to future flight test
TRACKING ANTENNA TEST PROCEDURE (HOUSTON) | ||
Procedure | Goals / Objectives | Knockoff criteria |
| Done in 15 minutes | |
|
| < list of reasons why you would want to stop the test > |
Move in periodic squares moving away from the tracking antenna. |
|
|
Flight Card 3
TEST PROCEDURE (PEGASUS 2) | ||
Procedure | Goals / Objectives | Knockoff criteria |
Preflight: Houston validation test with new motors | Ensure Houston flies stable with newly installed motors |
|
| Get GPS location of home position using your phone |
|
| Get a reliable ground truth value for the position of landing pad. Record the GPS location of this target landing pad as well |
|
| Please check manually that it is running for the first flight (on the ground before it takes off) using SSH or portable monitor | Airside system does not start, or it has not been running successfully before |
| Someone please check to make sure the drone is actually above it so we can get detections |
|
|
|
|
| Collect data for cluster estimation | Not enough time in flight test, or no more batteries Cluster estimation log or geolocation log is empty (no detections, if the detect_target_worker died bug shows up) |
|
|
|
Timeline
8:45 start charging 6S batts
9:30 All Attendees present in WARG Bay
9:45 briefing
10:20 leave the bay for the test
1
@Georgia Vachon Westerlund has records of timeline, will add when I get the photos off d800
Attendees
@Jerry Tian
@Georgia Vachon Westerlund
@Nathan Green
@Daniel Puratich
@Anthony Luo | Media | needs transport
@Ryan Scomazzon (Car if needed, wasn’t needed)
@Balaji Leninrajan Autonomy
@Tochi Okoro Autonomy
@Evan Janakievski
@Alison Thompson (car)
@Ben Lovegrove
@Andy Meng (car)
@Camilo Artigas Alos
@Smile Khatri
@Megan Spee
Key People Not Attending
If you’re directly involved in this stuff and cannot make this test please note it here. All onsite FTCs and subteam leads should fill this out please.
@Maxwell Lou
@Sam Zhang
@Omer Sajid
Debrief
Timeline
Departed Bay: 10:15am
Arrived: 10:40am
Debrief Begins: 1:45pm
Pegasus 2 Barrel Landing Attempt
good
called off when @Nathan Green didn’t think it would work
default tune was better than auto tune
bad
concerns about it clipping (contacting the drum and catching on the edge… or just hitting the edge a lot)
actual drum will have a lip similar to ours
task is significantly harder than anticipated
taking off from the drum results in a clipped edge
even if we make the drone smoother maybe not worth it
had to disable compass because the drum is metal
disabled Rm-3100 (external mounted to frame, not on poles)
drone drifts a lot from positionwe
check if this is due to tune or because of wind
Op flow was in the wrong orientation on monster mount → rotated properly and used tacky tape to reattach it
thoughts
landing drone in barrel is hard
it could be hella windy at comp
using acro, better tunes, there are ways to make stuff better
mounting the lidar and the op flow close together would be better – they are ‘tied together very closely’ in software (ardu).
for a dangling redesign, where pumps are below, it would only generate tension on the drone’s cable/line. so long as the pump is submerged and heavier than water with enough slack, the drone can loiter in wind gusts and have enough variable distance.
peggy 1 autotune? → went very well because we knew what params and behaviours peggy 1 needed. This peggy 2 tune is worse because we haven’t spent the time tuning. However the two frames are kiiiinda similar so might be able to use tune 1 on peggy 2. would be better to retune well obviously.
Speculation time
1-2 meters is within ground effect range.
occurred today with dropping the pump on a rope to the drum. loiter while the stick commanded down, built up a bunch of error and dropped suddenly. could have also been one of the lidars fucking up a bit.
Lidar stops being used over water (testing filling the tank) - stops over the water. reports a lot of distance. Likely switches to barometer. which is a bit more inaccurate. probably freaks out due to the change in height measured.
try turning off sensors and incrementally turn them on to check which ones are getting funny. figure out what causes the problems and what sensors improve what behaviours.
next steps
tune (water and no water) - autotune only tunes attitude
tune without water and run the drone with water to see what happens/how this affects things.
do an aggressive tune for position and default/regular tune for attitude
mech to fix op flow mount (monster mount)
monster mount on different arm than peggy 1?
offset is different from Peggy 2, correct the offset in Ardupilot
maybe mont the lidar on the monster mount as well?
could also attach the old op flow mount to another arm
can try props that don’t fold (will need to buy them)
this can be characterized by trying a polymer 22” propeller to compare the folding ones. compare the curves over a throttle up period see if they look diff over time.
use thrust testing rig? (never been used before)
Fly Houston to characterize the tune. run this next term: get people used to checking the logs, being able to diagnose, and tune well. this can be done on Houston for practice and then ideally this skillset is applied to peggy.
Pegasus 2 Water Dispensing Attempt
good
the camera worked to see the water stream and the bucket
didn’t help to be honest
home depot buckets
shoulda brought hammer but yea it’s fine
RC link and video seems fine on average
bad
leaks a lot gg
known issue with seal, fixable
the inlet was siphoning initially - the tube does stick into the tank by ~ 1cm
the angle of the tube can shift as of right now. not controlled by anything but like, random zip tied and duct tape. maybe make a mount.
RC failsafe’d for a second when the drone landed - as soon as the drone disarmed it said “RC failsafe cleared” - might be an ardu glitch
thoughts
tuning is the biggest thing for pegasus 2.
ideally the drone is tuned well enough that it can on god actually stay within 10 cm on each side. right now, it tweaks out really hard when it hits the bucket.
dangling thing should have a breakpoint….
if fishing line is connected, make sure that if it gets caught, it can break off- like the joints or the maybe fasteners that attach it are weak enough
the press fit tube fittings MIGHT be enough to be a target break point. however at an angle, they will not rip out easily.
“Drone doesn’t stay in the same spot when you let go of the controls”
opflow on white snow terrible. maybe generate some texture
would also have this issue on dirt or concrete or grass in alberta
“just be aware of this”
Houston Cluster Estimation Attempt
summary
rtl failsafe triggers at 10.5V for both attempts
log files were made for both, so there are images, but no detections were done
good
new motors are good (they stopped making bad/strange sounds) “no cooked”
bad
daniel discharging battery for analog video system on pegasus
autonomy script doesnt auto-start as it should
should be tested in the bay in the future
only charged 2 batts for Houston, did not establish in FT document how many batteries we needed for each flight test.
thoughts
how much current consumption does the new motors consume? Is it more or less than other motors?
about the same
Pre-heating batteries?
we have heating blanket with an XT60
could do in-flight heating
can we set the rtl voltage to be lower when we’re flying in the cold because the batteries will sag more when cold
could compare warm and cold batties
charging extra batteries to account for analog video system
we didnt plan how much we wanted to test so hard to predict how much we needed
forgot duct tape, hammer
have a procedure before leaving the bay and preparing for any flight test. list of items that should be in the flight test box etc. controllers charged. batteries allocated for each task.
used to have one but fell out of use. can go back and build off of that.
need to not forget stuff