...
This is the second ZeroPilot sync meeting in preparing for competition in early May 2023. The goal of this meeting is to set the minimum viable product from a controls and data perspective as well as the possible tiers of progression that are realistic before this upcoming competition.
The previous sync coverred covered all active competition efforts and is documented ->2023-03-14 Meeting Minutes - ZeroPilot Sync
...
Attendance
Topics
Minimum viable product assuming no autonomy points (Just aircraft flying)
Required subsystems, testing, and integration plan
If we wanted some autonomy points, how do we get them
The rules read through was on Monday so hopefully we have some idea here
list out subsystems and testing plan to correspond
Notes
Minimum Viable Product for competition 2023
Running Ardupilot as is currently used on Cornflakes
Mission planner replaces IMACS GUI
Tracking antenna replaced with running omni-antenna’s on tripods for IMACs
PPM out of RFD direct to Pixhawk Ardupilot
Mav-link direct to mission planner
More information regarding VTX specifically 3-21-2023 ICARUS Sync
More info found in Project ICARUS Weekly Sync as well
Everything denoted in this section has been tested in flight tests and will be pursued at competition
Minimum Step for flying ZeroPilot
Only advantage of doing this is we get more auxillary functions
fancy lighting control
fancy beepers
PPM passthrough would need to be tested and validated before ZeroPilot can be flown
Three or four flight tests are needed so this cannot happen
We have not flown ICARUS yet as well
Test PPM passthrough
Flight modes, decoding, & re-encoding logic
Test Mav-link
Houston Test
Fixed Wing Plane
Guided Modes
Cornflakes Test
Mav-link & PPM can both override each other so that can be used for testing one at a time
Mission planner and RC controller always has full control when using ZP passthrough
Next it to check telemetry and ground to mav-link and back
IMACs needs to be able to send and receive for this functionality to work
CV has no way to read codes and don’t have integration to path finding processes
There’s no guarantee we’d get the autonomy points because IMACS needs the integration
Doing this option opens us up to autonomous points (without autonomy having to rewrite drivers)
ZeroPilot will not be flown at competition 2023 for the above reasons
We will use mission planner
IMACs cannot send relative motion commands because mav-link does not support it. They might be able to send precise GPS coordinates, but of course this requires testing.
Jetson will consequently not be flown as well at competition 2023
ZP & Jetson goal is to fly at competition 2024 in the system planned as before
Autonomy integration to mission planner
This could be done to salvage our ability to get some autonomy points
This will be saved for IMACs sync for more details
IMACs sync will discuss landing pad detection
Jetson & ZP was planning on having a closed loop control system, Ardupilot is now the flight controller we plan on flying, we just lost the controls part of the controls system by dropping ZP, the only thing left autonomy can do is perception which can be done on the ground station using the incoming video feed from the pilot camera
Landing pad detection model still needs to be trained and code to run inference needs to be written. It is unlikely this will be ready
It may be worth saving the risk of crashing with the Jetson especially since a lot of this is untested
The system will have a lot of inaccuracies and is unlikely to be running at comp.
We will need to write new drivers as well to make this happen from a autonomy/cv perspective
Instead of taking time now to salvage autonomy points we keep autonomy on the current ZP architecture on the understanding it won’t fly at this comp 23, but next comp 24 we can make it happen.
Autonomy IMACs integration with mission planner will not be pursued for the above reasons
This leaves only other stuff CV is doing to be QR code scanning, path optimization, & other stuff which will be discussed in IMACs sync
mission planner can take in a list of waypoints, more details for IMACs sync
Possibly run dual telemetry for Ardupilot
Would mean IMACs team have two different outputs
We send one telem to ardupilot
Using dragonlink or LTE instead of RFDs, would be seperate from mission planner
this gives fancier light controls, easier to write waypoint path planning software integration
This will not be done because it offers negligible value for competition
Possibly run landing pad detection on pilot camera
If we can split the video we could possibly run landing pad detection on pilot camera
Landing pad detection could be done on ground station computer
This does not take away effort from people working on landing pad detection right now
“Pad selection = 10 pts”
We need to clarify if us putting a box around the landing pad counts
If this doesnt take time away from CV people (we need the code eventually) so this doesnt change too much
We would need to screen record the laptop and then include that in report sent to the judges
Laptop may have processing power limitations, do we bring the desktop to competition ?
This is a stretch goal primarily for autonomy that will be pursued due to possibility for more points at competition and doesn’t take away from longer term autonomy goals
Possibly Decorating the aircraft
Aidan Bowers (Deactivated) to discuss at ICARUS sync due to mechanical nature
Possibly LTE Telemetry
Would be a redundant telemetry system over cellular and not just radio
This allows us to RTL in case we lose RC link
Nathan & Anni have discussed this for a while (unsure on status in architecture doc)
We have Raspberry Pis in the bay to test with, Nathan will bring a cellular module for someone to debug with this weekend
To be discussed at general sync (or worst case ICARUS sync) as Anni would like to discuss with the entire comp travel team