Skip to content

Lab notebook

This is the working notebook for the project, copied from the repository as it stands. It is not polished. Every fact is meant to name its evidence, open questions are marked Q:, and wrong turns stay in with what replaced them. The only edits made on the way here are swapping the unit’s Wi-Fi name suffix for XXXXXX and checking nothing like a serial number or full MAC address slipped in.

Note What it is
Drone state What one Bebop 2 reported about itself at first contact
Protocol ARSDK3 wire facts, observed and read from code
Firmware 4.7.1 What is inside the official firmware image
First contact capture Every value from the first read-only session
pyparrot fork The patched library, each bug and its evidence
macgyverd The on-board crash reporter
Settings decisions Deliberate changes to the drone’s configuration
Lessons What went wrong, and what it taught us
Open questions What we still do not know
References Other projects and documents, assessed
Flight program and the motor board How the flight program talks to the motor board
Motor errors 8 and 9 What the stock flight layer does with them
Pad byte and tick length A motor board detail, read from the flat image
Session log The chronological record
Network Wi-Fi and addressing
ADB survey The drone’s debug shell
Phone app notes The raw notes behind the phone app page
Hardware reverse engineering The main hardware notes: buses, sensors, boards, with a verification ledger
Flight controller map An early map of the flight controller interface, partly superseded
Control law Loops, gains, limits and flight modes as read from the firmware
Mixer input roles What each input of the quad mixer controller is
Colibry sensor fusion The estimator: sensors fused and config values
Camera, ISP and vision The camera and vision chain in firmware 4.7.1
BLDC tools decompile Static analysis of the motor board test bench and host bootloader
BLDC update image The motor board update image, Bebop 2 against Bebop 1
Motor controller I2C protocol Draft: a first-pass protocol write-up, partly unverified
Motor board firmware write-up Draft: unverified and partly refuted, kept with its caveats
Flip acceptance What the flight program checks before it accepts a flip command
Lesser-known commands Animation and Follow_me commands the usual protocol XML leaves out
DemoMagicCarpet A demo feature in the flight program, and what is missing around it
Skycontroller NG firmware notes The NG model’s firmware image (not the Skycontroller 2)
Skycontroller NG stick path How sticks and buttons reach piloting commands in that image

Two different controllers are covered, and the notes say which. The first four were read from the system volume of a Skycontroller 2 (build mpp-linux-1.0.7). The rest were read from a firmware image that turned out to be the NG model’s (build nap-linux-2.0.0-rc3), which is not the Skycontroller 2’s. The services notes describe unauthenticated services as configured in the firmware, and each one says what was and was not checked on a running unit.

Note What it is
Skycontroller 2: hardware The hardware and firmware blobs in the system volume
Skycontroller 2: services and network Boot, services and network exposure
Skycontroller 2: software and mapping The software stack and mapping files, compared with the NG image
Skycontroller 2: button matching How pressed buttons become actions
Live unit against the NG image What a live Skycontroller 2 reports against what the NG image says
NG image: hardware What the NG image says about its own identity
NG image: software and protocols Its software stack and protocols
NG image: services and network Its boot, services and network exposure

Comments

Spotted a mistake, or tried this on your own Bebop? Leave a note. Comments are kept on this site's own server.