Please enable JavaScript.
Coggle requires JavaScript to display documents.
Px-4 vs Ardupilot software - Coggle Diagram
Px-4 vs Ardupilot software
Vehicle control
Stability
Mission planning
Mission critical device?
Power control
Telemetry
Vehicle position
Speed, health stats
Communication
Mavlink
(drone code foundation)
:warning: Drone CAN support
Support for other CAN node types needs to be checked
ESCs are supported
Don't want to be doing too much decision making in autopilot
Risks
Open source GPL3 licencing
Not well written for leasing
Licensing terms are not clear for non-distributed code like LGPL
Proprietary info
Needing to publish source code will impact business/ design decisions
Observable source code will give too much info on hardware
Need to talk to an IP lawyer for this
Pencil this in once we have money to service this
Perception of GPL
Businesses are allergic to it
Large amounts of in-house development of software
Cut over risk - big bang rewrites kill projects
Will likely need to rewrite everything again - not as plug and play as we hope
Rewrites have compound interest
:check: Use ardupilot for next 12 months and write code that will limit blast radius of switch
:!: Lets plan for the icky inevitable sunk time in rewriting code chunks
Bootloader
like for like with Ardupilot and pixhawk
Do we want to lock down bootloader with encryption?
Needs to be agnostic to firmware changes
Bootloader with too much capability
Feature creep and its a bit grot
Instead make a smarter low power controller board
Optimise later
Customisation
Custom modes
i.e. virtual anchor
Field programmable
Nice to have
Custom frames
Location, number, oritentation of thrusters
Switching frames at run time/ on the fly
PX-4
Poor support for underwater vehicles
Smaller user base
Often branched off into commerical products
:question: Roadmaps for major architecture changes
Younger product, learnt mistakes from Ardupilot
Better architecture - can add and remove components easier
Unknowns are stopping our decision - we need to time box this to make a more informed decision
Need to be evaluated against good criteria
More realistic way forward - but when to move is unclear
Reliability heirachy
Low power controller
Autopilot
BlueOS
Abstracting
:warning: Time box investigation into OS to unblock our hardware decisions
Want to move towards put a higher weighting on design time than asm time
Worth having a look at - it will inform technical decisions for the next prototype - benefit observed in next 6 months
Ardupilot
Better support and investment
Architecture summary
Low power controller smart
Bootloader dumb
Use Ardupilot bootloader
Autopilot
Ardupilot use for next 12 months- revist decision in 12 months
Investigation selection criteria - might come out from use
We know enough for a first pass
Make a list for next week
Understand PX-4 equivalent of CAN node