[Canberrauav] Minutes of CanberraUAV Meeting Monday 18th December 2018

james owens scoobydrive at hotmail.com
Tue Dec 19 02:43:08 AEDT 2017


G’day All,

I’m afraid the length of the minutes reflects the length of the meeting and the great quality of the discussion. It is probably easiest to view them here: https://docs.google.com/document/d/1L2AqLKgtJnHAaxgUf5aPCINMH8vTRMPdD-ql2qV8iAo/edit?usp=sharing

But if you can’t access that, then they are pasted below for easy access.

Please let me know if there are any omissions or inaccuracies.

Jimmy

Minutes of CanberraUAV Meeting Monday 18th December 2018


  *
Weekend flying
     *
CMAC. Greg demonstrated his HD goggles. Peter had a difficult to explain and significant crash in his Bixler; it was possibly caused by a burnt out ESC. Stephen flew his laptop and discussed some things with Matt and the SOAP Drones team.


  *
Intel Aero vs RasberryPi 3 - Stephen
     *
Stephen has been unable to get the IntelAero computer to recognise the it’s own cameras. Tridge thinks this may be a msg format issue.
     *
Stephen expressed a concern that Pi may not be able to process 10MP images for CUAV purposes.
     *
The 2012 CUAV image processing technique of two stage processing; first through reduced resolution scanning, then analysis of full images that meet the defined criteria. Tridge suggested that using a similar process might be a useful way of mitigating processing issues.
     *
Stephen and James discussed the possibility of trying an ‘Up’ board to see if that would offer better processing solutions. It was highlighted that we did not know if the latest ‘Up’ board has a relevant camera option.
     *
Tridge articulated a concern that we are too focussed on the image processing hardware; it was his opinion that the greater risk to the ‘Type 2 Autonomy’ challenge is the software. The extant CUAV software is dependent on human decision making based upon automated recommendations. In order to achieve the ‘Type 2 Autonomy’, the entire decision making process will need to be conducted autonomously.
     *
Tridge highlighted that the lack of useful representative image data for development and testing of the image processing software.
     *
Tridge and Stephen also highlighted that the size of the search area for UAVC18 is smaller than previous UAVCs; this will affect the image processing strategy.
     *
Dale and Darrell offered to fit PiCams to their platforms in order to conduct data capture for software development.
     *
Each collection mission will collect around 18GB of imagery. It was agreed that high capacity micro-SD cards would be required for the collecting PiCams; 64GB cards will need to be procured.
     *
Tridge highlighted that accurate time and geo-spatial data is required for effective analysis and geolocation. MavLink T-logs that correlate with images will need to be recorded and stored with the images. We will also need to record the ground-truth of the locations of the images for cross-reference.
     *
It was highlighted that we had not come to a satisfactory conclusion on the best target design.
     *
Reference was made to a VPN that has been established previously for remote software support for development. Zero tier may be useful for this.


  *
Decisions:
     *
Jimmy will write a trials plan for image capture to ensure that imagery is captured in the correct manner and documented appropriately for software development.
     *
CUAV will acquire enough Micro-SD cards to enable image capture (4?).
     *
Jack will print some example targets and mount them on corflute. They will be stored at CMAC when not in use.
     *
Dale and Darrell will conduct collection of representative image sets; they will need guidance on what is required to set up the PiCams appropriately. Stephen will flash the MicroSD cards with the necessary software.


  *
Pix4D Ground Spatial Distance calculations - Jimmy https://drive.google.com/open?id=11t-wnnlBtJctSrqfhwfIaeIFqvYpX8pakklvxi-cZ_g
     *
Jimmy highlighted the challenge that the geographic area covered by PiCam images captured at 120m would have for geo-location of targets to the accuracy required for landing within the competition parameters.
     *
Jack stated that we may need to look at a new strategy for refine the target geo-location based upon a descending altitude approach. Tridge suggested that this should be subject to further analysis pending development and testing of the CUAV software for UAVC18.


  *
Death Valley mapping - Darrell. Nothing to report.


  *
Should we run a UAV Challenge workshop in January? - Jimmy
     *
Jimmy suggested a CUAV workshop to share knowledge on the previous UAV Challenge missions and discuss the approach to the CUAV UAVC18 entry.
     *
Dale suggested that we open a workshop to the ‘MakeHackVoid’ membership.
     *
Tridge was hesitant to commit to a workshop, noting the time required to prepare for such activities, and the inefficient use of time for such activities based on the challenges of sorting individual computers. He also highlighted that the Mumble is a suitable place to discuss project planning details.
     *
Dale and Darrell stated that they would like to see a focussed ‘build day’ that could deal with discrete aspects of the UAVC18 project.
     *
Jack highlighted that we should make better use of Sundays at CMAC to make progress on the UAVC18 project. There was general agreement that it would be helpful to be able to work on UAVC activity through and after lunch.


Decisions:
- A task-focussed ‘build day’ will occur on 20 Jan 18, location TBC by Jimmy. The focus will probably be on imaging system build, fitting, set-up and image collection trial plan. Note that the following day is Sunday 21 Jan 18; it has the scope to be used as a follow-on ‘fly-day’ to capture lots of data together at CMAC.
- We should seek to identify work packages that can be prepared for prior to CMAC Sundays in order to use Sundays in a focussed way to achieve UAVC project development and testing  (and fun, obviously!). Mumble and the email conversations are a suitable place to do this co-ordination.


  *
Christmas Mumble? - Jimmy
  *
Decision:
     *
CUAV will take a break from Monday Mumbles due to the holiday season. The next Mumble will be on 8 Jan 18.


  *
Any other business
     *
Jack needs to buy props and batteries for the TVBS airframe -
Decision - Approved.

Notes

  *
Thoughts from Jimmy; should the image data collection activities include concurrent collection with the legacy Chameleon for a direct comparison to the PiCam capabilities?



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://canberrauav.org.au/pipermail/canberrauav/attachments/20171218/f2966b68/attachment-0001.html>


More information about the Canberrauav mailing list