2021-06-29 Competition Debrief 2021

Owned by Brielle Chenier
Last updated: 2024-04-13 by Daniel Puratich
4 min read

Slides: https://drive.google.com/file/d/1C7udN_QgiyowLNObOqV-z3Wez9_I--7b/view

The Good

  • Safety firmware and retreat-retreat worked well

  • Productivity and module testing took place at an impressive scale

  • We put a plane in the sky!

  • We won an award!

 

Telemetry

  • Got to developing it later so wasn't ready for competition

  • Video transmission didn't work with CV

  • Video transmission was only tested at Lucy's house

  • Only certain people were in the place/position to test - made a bottle neck so things had to be prioritized

In the future:

  • Integrate more in advance

  • Respond faster on status of telemetry task

  • More frequent videos of set-ups so they can easily be replicated by others

  • More range testing in advance - in real conditions, not just at someone's house

  • Test more

  • Treat it as a very important system (that it is)

Lessons learned:

  • Importance of frequent updates and good documentation that can be updated and can be used to replicating

  • Take photos of stuff such as connecting servos so that when someone else goes to do it, it's easier to reference. When work needs to be redone in the future, can easily get the same results

 

Zeropilot

  • Was never really tested, harder to evaluate

  • Wasn't ready for comp which is under stable since it is a big system

  • Huge flood of PR's close to comp, hard to review everything well

  • Members only got development boards in May

 In the future:

  • More frequent testing so code can be evaluated and improved

  • Better portfolio of boards to send to firmware members so that more people can test - need more dev boards

  • Better PR system, assign someone to improve the code

  • Better organization

 Lessons learned:

  • Testing boards should be prioritized

  • Develop layers for tests to occur - simulation layers should be developed so they can replicate the aircraft so we can test without boards

  • Hardware in the loop - need to check that code works for the board it was designed on, not just on a computer - every time a test is done flash the code to the board to make sure it works, hill test set-up

  • Continue to develop with simulations in Matlab - software in the loop

  • Need: hardware in the loop, software in the loop and unit testing!

 

CV - Architectural

  • Overly complex transmission system

  • Distributed system so failure potential between two computers

  • Data transmission was its own network

 In the future:

  • Possibly one system

  • Run CV on plane itself and have wired connection between 2 computers, not as complex as we originally thought

  • DGI uses Qualcom snap dragon (hard to buy), Jetson's use a low of power

 Lessons learned:

  • More steps = more failing points

 

CV - Operational

  • Generator fiasco, wanted to rent from Home Depot but then thought rocketry had one we could use

  • Ended up not having proper communication channels with rocketry

 In the future:

  • Responsibilities should be given in pairs, harder for lead to keep track of everything with a larger team

 Lessons learned:

  • Should think of back-up plans when close to comp, not relying on people/things out of our control

  • Keep track of tasks in Asana and person to reach out to

 

CV - Operational (Pt 2)

  • Tests were failing but couldn't replicate it

  • Poor repository management, used a lot of forks which was bad and taxi was only merged late into main

  • Unit tests were run but it wasn't run on comp machines

  • Tests were organized poorly

  • Documentation was lacking so debugging was slow

  • No easy way to test

  • Couldn't run test on production environment

 In the future:

  • Keep using branches for repo's

  • Should use CI Pipelines

  • Code base on production environment should have been reset to what's on the core repository

  • Test should be run in production environment before competition

  • Documentation should be completed external to code base (pdf/test doc somewhere), should have people in charge of documentation

  • CV create a docker image

  • Docker image could also be used for firmware

  • Make a better timeline - enough time for changes

  • Add a timeline to tickets (ex: geolocation took very long, target acquisition was very short)

 Lessons learned:

  • Lots of timeline issues and documentation

  • Prepare for a "dress-rehearsal" a week before comp to minimize last minute tasks

  • Integration testing should start a lot earlier - as soon as 2 models are done

  • Full testing rig

  • Asana was really helpful! (maybe we could get Jira)

  • Dual responsibility

  • Have a minimum viable product - something a lot simpler but we know will work

 Moral of the Story:

  • Complex system with not enough time to test

 

Flight Testing

  • WARG wasn't able to flight test anywhere near enough that we should have

  • Only Anthony is able to fly Spike

  • Not enough airframes, we shouldn't have spent time fixing airframes at critical moments of the year

 In the future:

  • Have lots of airframes in advance - "infinite supply"

  • Have more RC pilots - a long process that takes a few months so hard to train enough people

  • Could recruit RC test pilots - clubs in KW, aviation majors

  • Thoughts should be put into how the test pilot team will be allocated

 Lessons learned:

  • More pilots

 

Gimbal

  • Expected fixed-range servos - we got free rotating ones

  • Prototyping should have been done a lot earlier!

  • Didn’t have a full CAD model so we're going based on measurements that aren't the most accurate

 In the future:

  • Test plan for all mechanisms

  • More documentation so other sub teams can see what we're doing

  • Have a full model of a plane to see changes

  • Have a solid inventory - we don't know what we have in the bay for any sub team

 Lessons learned:

  • Focus members on this year's competition

  • Prototype a lot more and make iterations as needed

  • Design model with the complete plane CAD to make sure it fits/clearances and all that

 Other:

  • Make a single axis gimbal to test CV

 

Grabber

  • Hit the ground and snapped before flying

  • Was tested indoors on smooth floor, should have been improved early on

 In the future:

  • Develop test fixtures on a bench so don't have to test with plane

  • Should have focused off-site members to redesign/improve

  • Prototype early on

 

Communication

  • Take every opportunity possible to test and use flight tests to their full extent

 In the future:

  • Have someone from every sub team at competition

  • Someone should be checking the competition manual and acting as a PM

  • Have some protocol between inter-sub team testing

  • Go more to other sub team meetings

 

 

 

 

 

 

 

Â