A place for West Australian UAV / UAS enthusiasts or businesses to discuss topics, arrange meets or share experiences.
Hexicopter
Hi AllPurchased a FlyPro X600 and I am trying to sort out what would be the most suitable camera to use for FPV and some video/stills photography. The supplier has wired it up for a GoPro 3. That model is a little out of date but a good camera to start with although if I am going to purchase a camera I would like to only spend on a one of that last for sometime. I have had a look at the specs for the for what I will be flying it for Sony FDR-X1000V. The software has stabilisation and some…
Read more…
Comments
@JB - I think the Silvertone airframe is way overpriced however I was posting it as an example of what we could do in a low drag airframe.
As to the resolution why not reserve judgement until we get some cameras flying. A human may be able to spot Joe with lower simpler resolution and we forgo the 10 points for a simpler visual system.
I found these http://www.advmw.com/uav.html. I have all the prices. The COFDM modules are about 8 to 10k however the analog video modules are in the price range of 1 to 2k. The main advantage of these modules is that they are rugged and designed for operational use. I believe we have to think of what equipment we use as if we are taking it into combat like the military do. These modules can be ordered on the Amateur Radio 1.2G frequency and I could pretty quickly have an amateur licence to make it legal.
@Hai Tran - Completely agree about the airframe. We should plan to recover from a crash. A simple rugged airframe with lots of spares will help. I am thinking a plywood or fibreglass fuselage with foam wings either vac-bagged or balsa covered.
Does anyone know if CyberTech could be approached for sponsorship?
Coptercam HQ is Unit 3, 289 Camboon Road Malaga, behind the Ultratune on the Corner of Beach Road.
You guys are in luck, PerthUAV.org is owned by.....me and its not being used for anything.
From memory the search area is about 2nm x 3nm with a 1nm transit from the airport.
A good read is this document University of Adelaide OBC
A few things I noted over the last 3 years of following the OBC.
1. The UAV might crash, so getting an expensive airframe might not be a good idea.
2. University of Adelaide published some info on a search logarithm for auto camera search
3. There is alot of documentation required
4. I have a nanostation and whilst it's good with a teradek, the bit rate is low. What is needed is an expesnive Cobham COFDM system!, I think the cost is upto $100k.
5. Mugin has been proven at OBC. Although if it has same design as Hugin, transportation might be an issue with the booms fixed to the wings? I have a hugin that isn't built because it doesn't seem transportable.
Now not trying to be negative, but Coptercam was a result of trying to get an OBC team and failing, it was far easier to setup a UAV business than to get an OBC team together! If we end up building a UAV that can do the job, it might be better off being commercialised rather than all the IP being released and published for a very small reward compared to the time and resources put in.
PerthUAV is already taken I'm afraid.
I don't mind the silvertone, but I can't see a decent place to mount the cameras/comms. Besides, for $18 000 its a bit steep for what she is...even though she's a looker! I'd rather spend that making our own molds. You could nearly get ten Mugin airframes for that. I like the ten hours of endurance though...only possible with fuel sadly enough, and not really a requirement for the 60min OBC course.
Just to give an idea how BIG the search area is, on the basis that we are looking at around a 10sqkm search area, if we do 200m wide search paths per run, we'd need to be averaging over 60kmh (16m/s) to cover that area in 60 minutes (we'd need over 60km range too!). If we double the search width to 400m (which is maybe too wide for resolution purposes I think), or if we double our speed to 120kmh, we will need 30 minutes to cover the same area.
I think we will need to aim for a 30min full search area cover, if we want to be the first to drop the package. That means something like a 90kmh scan and nearly 300m wide path!!
Just to compare: James overhead photo is only about 30m wide...
A single 5MP camera will have about 12cm resolution at 300m wide. So a body will only be about 4x15 pixels! James photo is about 3cm resolution (if his camera is a 540TVL) in comparison. We might be able to use the Raspi with 5MP camera module: http://www.raspberrypi.org/archives/2555
It also has onboard encoding, and we might be able to get it to do some onboard algorithms. LAN is standard on the Raspi at least. For the same resolution as James video we'd need to run at least 4 of them, and then stitch them to avoid to much overlap (it would be nigh impossible to hold the altitude accurate enough to avoid auto stitching), and then somehow live view or autofind Joe, all whilst the UAV is moving at some 90kmh! The downlink would have to be able to sustain +20MP photos every 10 seconds or so, which is about 2MB/s for RAW or 500kB/s for JPEG. Achievable with a good link I suppose. At least we can go up to 1500ft to get the range when we're 10km out from base.
Video with the 5MP cameras running at 1080p instead would result in a 2.5 time reduction in resolution, so we'd need more cameras, say around 10, but the data throughput would need to be around 200Mbps! So pictures with auto stitching it is I think!
Well they did say it was a "challenge"!! ;)
Can we just be PerthUAV? Anyway if we are accepting sponsorships we perhaps need to form something to handle the money. We do need a dedicated forum.
Anyway baring accidents I am good for Sunday.
The advantage of our own airframe is that we can use low drag airfoils. Twenty knot winds need a high wing loading and low drag to be stable. We have to fly in these conditions. Some of the designs out there are not low drag. Mind you the Mugin of CanberraUAV seemed to handle it OK.
Perhaps something like the Flamingo:
http://www.silvertone.com.au/silvertone.pdf
This airframe is $18 000.00 to buy.
Coptercam 10am - all good
can you email the address?
John. I was thinkiing of using the Nanobridge M5-25 on the ground with tracking (i have a few spare) and the Nanostation Loco M5 on the plane, possibly even with some sort of limited tracking to help overcome the polarization problems. The max range on OBC is about 10km, so they should be good enough as long as we have enough altitude, hence the camera type question. High and slow but with a high resolution coverage would seem to be the best way to go. At least the down link will be good then when the aircraft is stable. I suppose we're making a glorified remote controlled flying scanning machine...
BTW on manufacturing:
I'm in the process of getting a Makerbot 2X 3D Printer, and also have a lathe, vacuum fiberglass gear&table and welding etc tools. Not that i expect us to do much welding!
I can do the CAD/Inventor/PCB design as well for any custom builds we need. I also have some skill in energy management/profiling/analysis, and structural FEA calculations, and even some fluid dynamics for putting together our own air frames if necessary. ATM I think we really need about +5 guys to be able to share the load and pull this off. I can setup a website/server/dedicated forum as well once we commit to streamline development. Do we need to officially form a club, or are we all going to be operating on a handshake? Personally I'd prefer a handshake...and only go down the club path should we ever have enough interest in membership.
On names: Didn't someone mention Western Skywalker or something? Maybe something shorter would be better like: Robosky or SARsky. It should really capture the automated SAR UAV theme somehow. sarsky.com is available atm.
What sort of range is the maximum used at the OBC? I've had a play with Ubiquiti Bullets (5 - 6Ghz) and the range wasn't brilliant... It was tested indoors though using a cloverleaf / helical combination. What models were you thinking of using JB? I've heard Nanostations are quite good?
Lol James
Hey i was just trying to give you a moral boost as you're likely to have much of the work load cut out just for you! Programming isn't my strong suit...yet!
Well FLIR would be nice, but has Joe just got a lamp or a flashing beacon? If it's just a lamp, and the frequency is unknown, it will be hard even with a FLIR. They tend to have narrow field of view and low resolution to boot, so maybe not that ideal for a daytime SAR. I think a FLIR will be mostly out of budget as well.
I was thinking along the lines of a 4 or 6 camera array and video stitching or high frame rate "panoramic" pictures. It will all depend on how fast the cameras are; the faster we can scan, the faster the aircraft can cover more ground. The higher the resolution, the higher the altitude on a single pass with enough detail to spot Joe. Some basic optic calcs would be the first thing to do, and finding the appropriate cameras. Any camera suggestions?
James your picture shows just how sensitive the whole searching side is.
On that basis I think that ground based image processing would be the go, but that requires a stable high throughput data link from the aircraft. Especially so if we are running multiple HD cameras. Have you already done some video search algorithms in software then? That would be a cool head start!
Alternatively, there might be a possibility to even make up some of our own camera hardware running on Android/Linux installed on the aircraft as well. There's a few development boards from Freescale/STM and the like that are already available for 360 degree car views etc using multiple cameras and real time video/image stitching. They run upwards of 1GHz quadcores with all the video compression etc onboard and can operate from batteries too. If we can get the search algorithms down to a tee onboard, we might be able to get a fully autonomous UAV, and avoid any issues from the down-link side.
I can come up on Sunday as well to Malaga for a meet. Some time in the afternoon would be best. We can still work on the "concept" until then as well. We really need to have the OBC rules as well, otherwise it'll be hard to scope whats needed and whats not.
As for the IP and methods being public domain, meeting the documentation requirements doesn't necessarily mean complete disclosure. Many designs are public domain, but things like programming and specialized air frames or electronics need not be all divulged in repeatable detail.
Typically I am actually all for "Open Source" under a creative commons licence, especially in the software development side. There's just so many hours that can go into getting functions to work properly, and even programming new ones, that it's the best (and nearly only) way to stay on top of the innovation curve.
The competitiveness of open source is also very good. The more ideas the better really, provided the framework supporting all of that is "structurally sound". Look at Android development etc. Most of the functionality is community grown, it doesn't make sense to restrict it by locking up the IP. I see current development becoming the reverse of the 80-90's where software manufactures made the money, i think in the short term hardware is going to be the enabler, and the money maker. Software is on it's decline, because it's just to silly competitive and so much can be easily outsourced now as well. We need to use that to our advantage.
John your 3G setup sounds good for a backup setup, but I think i'd prefer the 300Mbps of the Ubiquiti setup at that range instead, if we can really get it to work reliably. Polarized antennas and banking planes don't always agree with each other! I suppose if we have an autonomous version with onboard video processing it will work though. you can never have enough options!
My property is here: 34°31'48.67"S 117°40'31.96"E
Between the two dams and extending from the cul de sac road to the back fencline. You can see the fence that divides the block in half on Google Earth.
The area is cleared and there is a clear view all the way to the ranges where there are sheep, hay farms and very few people. That's why I bought it ;-) .
Probably not suited to long range testing... it would be hard to follow by car! But if you get desperate for space... go for it.
As you know I have the Mugin sitting around and was to be my testbed for my NextG streaming gear. Maybe you could convince me to "lend" it to you if you decide to go down that road? I've got the 55c engine, electronic ignition and fuel tank, but no servos as yet. I would love to be involved with the OBC (it's been a dream of mine for years) but I will only be able to help in short bursts due to my work commitments (at least until November)
@JB I think video would be cost effective, especially if we run an IP cam that can have it's sttings changed on the "fly..." Off or low res for launch and transit to target area.... then High res (and whatever fram rate you choose) on target. It's supposedly good for 7.2Mbps uplink but who knows what it will actually acheive in real life? Data packs are chosen on a month by month basis so can be topped up when needed for testing. It has two serial ports RS232 level (but i've had them working at TTL levels using a converter) and two lan ports.
I also have a lathe if we need custom parts, spacers, mounts etc.
-
3
-
4
-
5
-
6
-
7
of 31 Next