[Canberrauav] CanberraUAV Meeting Monday 05 Feb 2018 - Minutes

james owens scoobydrive at hotmail.com
Mon Feb 5 21:07:58 AEDT 2018


G’day all!

The minutes, with pictures, are located here:

https://docs.google.com/document/d/1gXEYJLqsh7DgsmemQ-RtBlKyVN0iCp0MRSrjoEsPWXM/edit?usp=sharing

And pasted below for more comfortable viewing on the personal electronic device of your choice:

1. Minutes from last week:
https://docs.google.com/document/d/1gqJv3Ee0BjxIaRirRsjHOfkbyAQ7-gudlZl08W8766o/edit?usp=sharing

2. Actions from Last Meeting

Action item: Jimmy to send this letter (with James' input) - Closed
Action item: Dale to draw up a 3D mount that the camera and lead weight can be mounted
Action Item: Elizabeth Lee MLA visit - Successful - closed

Agenda

3. Last Week Flying -
Jack flew the Boomerang for around 15 mins - it seems to make fuel!

4. Imagery Collection
Saturday evening Tridge added 50g lead weight to the camera for the Ranger in a basic hot glue mod. Jack put in a temporary wind shield onto the Ranger too. The impact on the quality of the image that was subsequently collected on Sunday was very significant. This significantly reduced the blurring on the image. We now need to collect in different light situations.

Stephen has separated the Chameleon parts of the software from the CUAV parts and it is all now Python 3 compatible. Next phase is to pull out the Chameleon stuff from the core code so that it can run on the Raspberry Pis.

Action: Collection in different light is now a requirement; cloudy days at CMAC, mornings or evenings provide options. We can conduct concurrent activity with the software in the mean time.

Jimmy; is there a requirement for RTK to increase our accuracy? Tridge; no, not really. The difference in the GPS error between collection and landing is the delta in our numbers. The change in a 5-10min period is not significant to the relative position between the Ac and target location is probably 1-2m. This may be a stretch goal, not a core requirement. James said that we are unlikely to be able to survey the ground station into provide significant accuracy.

Greg sent out some very useful info on ISO settings last week, which is repeated here:
“On the mumble tonight we discussed ways to improve the pi Cam Mount.  I mentioned you can use the 4 holes in the Pi Cam v2 Pc board and nylon standoffs to any type of plate you want weighted as appropriate. Simple and no casting Pb.
I have attached a couple of photos of this screw mounting.  I bought this out to Cmac back in November but it seemed to generate no interest.  I did some photography of our targets on a short auto mission with the Pi Cam mounted on my Solo at the time and posted a link to google drive.
https://drive.google.com/folderview?id=1nyWOgQVejr750ZsSgPaDy5I8P0kVY-ri
The are also 2 gopro images ending in 188 & 199 for comparison.
The day was overcast and as we also  discussed ISO on the Mumble these images have the EXIF data.  ISO from the pi Cam was 50 so fine grain + low sensitivity and speed seems to vary between 1/1000 and 1/1200 of a second. (see attached)
Greg”

5. The Kraken! Greg started the build at the Make Hack Void build day. Greg’s build images are here: https://docs.google.com/document/d/1gXEYJLqsh7DgsmemQ-RtBlKyVN0iCp0MRSrjoEsPWXM/edit?usp=sharing
. The take-off run was about 2m! Need to do some CofG adjustment and will be adding the PixHawk set-up this week. Jack asked if the aircraft would lift itself vertically; this may be worth trying. Jack provided advice on how to add servos for a TVBS setup for the wing.
https://youtu.be/Qy8_3kKvTFw

6. Any other business?

7. Greg. For Tridge: with the image sets that we have is it worth looking at a Deep Learning code like Tensa Flow. Trige: it would require a significantly different hardware set-up. Greg can access the datasets so that he can try this if required. Tridge cautioned that the hardware available on the ground may not be achievable in the UAV. James. TensaFlow can run at 1-2Hz and a reduced version can be used that is called ‘Mobile Net’, but this is probably not going to be as effective as the current CUAV set-up. A cut down version of the image processing from CUAV and something like inception may work. General consensus was that this option is probably not viable.

8. Fiona Spear may come to CMAC on Sunday 11 Feb to discuss the following with us:
“I am from the Southern ACT Catchment Group, a Not For Profit organisation that supports landcare and parkcare groups in the Woden, Weston, Tuggeranong areas and further out to Paddys River and the Naas areas.

We are looking at running a workshop for local landholders to show them the sorts of things that drones can help with on there property. We aren’t looking for how to fly them, rather just a taster of what they can do, so that individuals feel more confident about their potential, and maybe then go out and learn how to fly them.

I was wondering if your club run such workshops, or if anybody would be interested in doing so? We do have some funding, but as a Landcare network we are limited.”

Action: Jimmy to confirm visit and lead hosting. All welcome to contribute or offer solutions. Darrell has already received a copy of the email.





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://canberrauav.org.au/pipermail/canberrauav/attachments/20180205/c39ca395/attachment-0001.html>


More information about the Canberrauav mailing list