Dave's Canyon Quadcopter Idea

[EEVblog]

Split off from another thread I didn't mean to hijack.

1) just use the main rotors for lift and maximum stability (on a very wide arm base), and use two additional propellers to push the copter forward. I'm guessing this would give it a much better stable video platform.

Actually that becomes quite difficult because you'll need a reversing brushless motor controller in order to stop.  Also, you'll still have to provide a mechanism for tilting the copter in order to turn.  Simply yawing by using two forward thrust motors and differential control will see your quad skating sideways instead of changing direction very well -- a bit like a hovercraft is very poor at negotiating turns.

That might actually turn out to be rather desirable?
I want the effect of the camera kind of "floating" through the canyon instead of obviously "flying" though with great precision.

A better way is to do what the Chinese do with the T580 quad and that's to use the accelerometers and gyros to not only provide a stable platform but to also automatically tilt the camera platform to compensate for any tilting of the quad itself.  If you tilt the quad forwards to travel forwards, the stabilized camera platform automatically tilts up so as to ensure the camera remains horizontal.

Yeah, I've seen that. My idea was to just try and eliminate that and make it simpler, because I don't have the same performance requirements as regular quadcopters.
I need it to do just one specific task, get one continuous shot through a complex canyon.

2) Fully autonomous, able to maintain a (rough) height of several meters above the ground, and away from side walls.

That's also been done.  Even the cheap T580 has a barometric altitude hold capability and the FyTech 91Q has inbuilt GPS and magnetometer to provide directional hold and position hold.  As for staying away from walls -- there are some cheap Chinese hovering toys that do this too, using either ultrasonics or LED proximity detectors.

Yeah, but the devil will be in the detail I'm sure.
Canyons are so narrow (several meters to say 30m wide) with a complex terrain that generally drops (via waterfalls), with all sorts of fallen logs and vegetation on the walls etc.
No GPS reception at all.

I originally thought to just fly a regulator stabilised quadcopter through, but the twisting nature of canyons means you have to "hand-off" control to several people stationed along the way many times as the canyon changes direction and you lose line of sight control. And the complex terrain floor (sometimes with waterfalls you have to jump or abseil, means a single controller following the copter through would not be possible.

I'm probably best just experimenting with a regular stabilised video capable quadcopter first (suitably modified for canyon wall impact and water landing protection) and see what problems actually need to be solved first!

Anyone got any recommendations for a low-ish cost stable video platform system that could be modified like that?

3) Have large foam floats on the base for possible water landings

Haven't seen that yet.

It's for "unanticipated" landings 
Actually the terrain floor could be anything. Water, huge boulder fields, sand, logs etc.

The goal is to have it autonomously navigate a canyon (and film it). hence no GPS reception, little radio reception, and basically start it at the entrance to the canyon and have to come out the other end intact, without having to follow it on the ground (which is hard and maybe dangerous)

Now you're getting into cruise-missile territory with terrain-following.  Uncle-Sam will kick your arse if he catches you doing that, even though you're an Aussie.  Look at what they did to me for building a cruise missile in my garage!

[/quote]

Sounds like fun! 

Dave.

[Trigger]

When I was working on UUVs (unmanned underwater vehicles) in the navy we had a similar problem.  How does an autonomous vehicle move over unknown bottom contours along a course and not run into anything.  We ended up using a high resolution short range sonar system.

With an aerial platform weight is an issue so whatever the system is needs to be light and 'good enough' to do the job.  Radar would be the typical answer but that's too expensive.  I'd try playing with some strategically placed laser rangefinders with an ultrasonic backup for altitude.  A small scanning ladar would be ideal but getting it small and light enough will be a problem.

[RCMR]

If I had to solve this problem I'd build *two* quadcopters.

One would fly a path above the top of the canyon (where it would have clear GPS and could thus fly a precise path) and the other would fly through the canyon.

I'd use a video link from the lower quad to the upper quad and then back to the pilot station.  Likewise the RC signal would travel from the pilot station to the upper quad which would relay it to the canyon quad.

This way, you could fly FPV (first person view) through the entire canyon without the risk of losing the RC link or the video link necessary for guidance.

Also, some prudent use of frequencies would help mitigate the effects of terrain.

Dave, I don't know if you've watched this video:



it's not canyon flying but not too far off it  and shows what can be done with the right gear and skills.

[Trigger]

If I had to solve this problem I'd build *two* quadcopters.

One would fly a path above the top of the canyon (where it would have clear GPS and could thus fly a precise path) and the other would fly through the canyon.

I'd use a video link from the lower quad to the upper quad and then back to the pilot station.  Likewise the RC signal would travel from the pilot station to the upper quad which would relay it to the canyon quad.

This way, you could fly FPV (first person view) through the entire canyon without the risk of losing the RC link or the video link necessary for guidance.

Also, some prudent use of frequencies would help mitigate the effects of terrain.

Dave, I don't know if you've watched this video:


it's not canyon flying but not too far off it  and shows what can be done with the right gear and skills.

That would actually be fairly effective.  You use one quad copter to relay the control signal and video for the other.  The high side would have to be bigger but all it would need to do is keep line of sight with the canyon flier.  That saves a lot of money for sensors and lost prototypes testing the autonomous terrain following.

There was a presentation at DEFCON last year for a DIY UAV.  It would be easy enough to build one that would loiter over and follow the canyon drone.

http://youtu.be/Dim2-rEO9j4

[EEVblog]

If I had to solve this problem I'd build *two* quadcopters.

One would fly a path above the top of the canyon (where it would have clear GPS and could thus fly a precise path) and the other would fly through the canyon.

I'd use a video link from the lower quad to the upper quad and then back to the pilot station.  Likewise the RC signal would travel from the pilot station to the upper quad which would relay it to the canyon quad.

This way, you could fly FPV (first person view) through the entire canyon without the risk of losing the RC link or the video link necessary for guidance.

Also, some prudent use of frequencies would help mitigate the effects of terrain.

Dave, I don't know if you've watched this video:



it's not canyon flying but not too far off it  and shows what can be done with the right gear and skills.

Our sandstone slot canyons in Sydney don't have straight walls and open skies, they often close in above themselves (multiple times curves) like caves.
We have major problems using 0.5W UHF walkie talkies in canyons.
Maybe some sort of strategic placement of RF repeaters pre-placed throughout the canyon?

Dave.

[Trigger]

If I had to solve this problem I'd build *two* quadcopters.

One would fly a path above the top of the canyon (where it would have clear GPS and could thus fly a precise path) and the other would fly through the canyon.

I'd use a video link from the lower quad to the upper quad and then back to the pilot station.  Likewise the RC signal would travel from the pilot station to the upper quad which would relay it to the canyon quad.

This way, you could fly FPV (first person view) through the entire canyon without the risk of losing the RC link or the video link necessary for guidance.

Also, some prudent use of frequencies would help mitigate the effects of terrain.

Dave, I don't know if you've watched this video:



it's not canyon flying but not too far off it  and shows what can be done with the right gear and skills.

Our sandstone slot canyons in Sydney don't have straight walls and open skies, they often close in above themselves (multiple times curves) like caves.
We have major problems using 0.5W UHF walkie talkies in canyons.
Maybe some sort of strategic placement of RF repeaters pre-placed throughout the canyon?

Dave.

The problem with too many repeaters is control latency.  You have the latency of the video getting to you added to your reaction time plus your control input getting back to the unit.  You'll have to test to see how much you can tolerate at the speeds you want to go.  A multiple radio mesh network might work, one radio mesh units will likely introduce too much lag.

[Psi]

Just a FYI, in case anyone is pondering buying a Parrot ARDrone, be aware that they had/have some code bugs where they ignore control input and fly up up and away (never to return once they catch the wind).

Had a friend who lost his ARDrone this way. It's apparently an issue with them that's more common than they will admit.

[EEVblog]

The problem with too many repeaters is control latency.  You have the latency of the video getting to you added to your reaction time plus your control input getting back to the unit.  You'll have to test to see how much you can tolerate at the speeds you want to go.  A multiple radio mesh network might work, one radio mesh units will likely introduce too much lag.

Speeds would be very slow, almost walking pace.

Ideally I'd use one of those RC helium blimps!
But the logistics of lifting a suitable camera is just a tad too high.

Dave.

[Uncle Vernon]

Our sandstone slot canyons in Sydney don't have straight walls and open skies, they often close in above themselves (multiple times curves) like caves.
We have major problems using 0.5W UHF walkie talkies in canyons.
Maybe some sort of strategic placement of RF repeaters pre-placed throughout the canyon?

Dave.

If you can clamber all over the place surveying and installing RF repeaters then wouldn't you takeyour photos/video while doing that? Why the quadcopter other than the challenge?

[EEVblog]

If you can clamber all over the place surveying and installing RF repeaters then wouldn't you takeyour photos/video while doing that? Why the quadcopter other than the challenge?

A mate and I had the idea one day while in a canyon doing photography and some video:
"Wouldn't it be cool to get one continuous video shot flying through the canyon?"
Sure, you can hand carry a video camera through the canyon, but it's just wouldn't be the same.
To our knowledge no one in the world has ever done it. Very occasionally you'll get a pro film crew in for a doco or something and fly a camera along an installed guideline or something, but it's only one part of the canyon.

So yeah, it's the challenge of it, because we can...

So anyone got any ideas on what the best platform to start with would be?

Given the tight nature of the canyon and obstacles, I'd imagine an enclosed rotor (like on the Parrot AR Drone) would be essential to prevent accidental bumps against the walls or rock or logs etc to not crash the thing.

Dave.

[IanB]

I think I would try to rig up some kind of dirigible. You would get longer flying time, greater resilience against crashing and a simpler control and navigation problem as you wouldn't be worrying so much about keeping the aircraft stable and level.

[EEVblog]

I think I would try to rig up some kind of dirigible. You would get longer flying time, greater resilience against crashing and a simpler control and navigation problem as you wouldn't be worrying so much about keeping the aircraft stable and level.

That was my first thought, but I calculated, IIRC, you would need roughly a 600mm dia helium balloon to lift 100g.

Dave.

[Trigger]

I think I would try to rig up some kind of dirigible. You would get longer flying time, greater resilience against crashing and a simpler control and navigation problem as you wouldn't be worrying so much about keeping the aircraft stable and level.

That was my first thought, but I calculated, IIRC, you would need roughly a 600mm dia helium balloon to lift 100g.

Dave.

Not to mention how much wind would affect it's control.  It was also my first thought but they have to be rather large to be a practical platform.

The relay system would be the most practical to build.  The biggest difficulty would be creating the TCP/IP based control system that would tolerate routing and roaming among access points.

[KuchateK]

The biggest difficulty would be creating the TCP/IP based control system that would tolerate routing and roaming among access points.

I've built WiFi networks in very large multi-room warehouses (with thick concrete and sheet metal walls, plenty of signal loss and interferences from neighborhood networks). You can have bunch of access points in the area and clients will jump between them as you move without any big issues (single packets are lost during jump). It is very seamless for the clients. Even though people told that simple won't work, no other fancier methods offered such reliability during testing.

I've also worked with city wide WiFi. We decided to drop "repeaters" around. One AP was used as client with highly directional antenna to the base and another provided coverage for the area.

Considering UAV based on Paparazzi (linux on beagle board) that would have power to handle TCP/IP and WiFi it should be pretty cheap and easy to build mesh network for large area. Old equipment can be acquired for free (B should be plenty) and it is very easy to get good antennas for it. And you can connect each AP directly to base station to minimize lag.

Another idea is to get balloon mentioned earlier or second UAV. It can be sent way up and work as a base station/repeater for much larger area. It can also move together above UAV invisible directly for the operator. If you want to fly below something use local repeater.

[KuchateK]

That was my first thought, but I calculated, IIRC, you would need roughly a 600mm dia helium balloon to lift 100g.

Lift goes up much faster than the size of balloon. Going from 2 ft. (~60cm) to 4 ft. (120cm) doubles the diameter, but lift is multiplied almost by 10.
Check the table here: http://www.chem.hawaii.edu/uham/lift.html

[mikeselectricstuff]

If you can clamber all over the place surveying and installing RF repeaters then wouldn't you takeyour photos/video while doing that? Why the quadcopter other than the challenge?

To our knowledge no one in the world has ever done it. Very occasionally you'll get a pro film crew in for a doco or something and fly a camera along an installed guideline or something, but it's only one part of the canyon.

If you can pull it off, and get good quality video, I'd imagine the footage (and making-of) would probably have enough commercial value to at least pay for the gear.
Slow definitely feels like the way to go, as it gives you time to react, and reduces constraints on live video bandwidth. Also makes for a wider range of possible playback speeds if you acquire a shitload of frames, and probably makes post-production stabilisation more possible. 
Helium (or helium asisted) may be a good way to go if you have the width clearance for it, as you save a ton of power for the lifting - assuming wind isn't too much of a problem.

[mikeselectricstuff]

I think I would try to rig up some kind of dirigible. You would get longer flying time, greater resilience against crashing and a simpler control and navigation problem as you wouldn't be worrying so much about keeping the aircraft stable and level.

That was my first thought, but I calculated, IIRC, you would need roughly a 600mm dia helium balloon to lift 100g.

Dave.

Remember balloons don't need to be spherical. I'd imagine a sausage or blimp shape may give you a better aspect ratio for flying a canyon.
 

[mikeselectricstuff]

I think I would try to rig up some kind of dirigible. You would get longer flying time, greater resilience against crashing and a simpler control and navigation problem as you wouldn't be worrying so much about keeping the aircraft stable and level.

That was my first thought, but I calculated, IIRC, you would need roughly a 600mm dia helium balloon to lift 100g.

Dave.

Remember balloons don't need to be spherical. I'd imagine a sausage or blimp shape may give you a better aspect ratio for flying a canyon.
 

We have major problems using 0.5W UHF walkie talkies in canyons.

For control you could go way lower in frequency, and with a bimp you'd have plenty of length for an antenna.

[ejeffrey]

For control maybe, but if you are flying this think RC, don't you need a video feed?

[EEVblog]

For control maybe, but if you are flying this think RC, don't you need a video feed?

yes, I'll need a live FPV video feed to control it, as the canyons twist and turn.

With the blimp, it doesn't matter the shape, it's the fact that you have to lift X weight, and you have to transport and inflate on site in remote locations.

Winds can be a problem in some canyons.

Dave.

[EEVblog]

That was my first thought, but I calculated, IIRC, you would need roughly a 600mm dia helium balloon to lift 100g.

Lift goes up much faster than the size of balloon. Going from 2 ft. (~60cm) to 4 ft. (120cm) doubles the diameter, but lift is multiplied almost by 10.
Check the table here: http://www.chem.hawaii.edu/uham/lift.html

Based on these practical one you can buy:
http://www.southernballoonworks.com/rc-blimps/remote-control-blimps.html?gclid=CKuC5cXDtKoCFYIlpAodAkij5A
A 2.4m long helium RC blimp can only lift 170g!

Dave.

[KuchateK]

Based on these practical one you can buy:
http://www.southernballoonworks.com/rc-blimps/remote-control-blimps.html?gclid=CKuC5cXDtKoCFYIlpAodAkij5A
A 2.4m long helium RC blimp can only lift 170g!

This might be the bad example. They are probably made from heavy material to last transportation between events and technical skills of marketing department

6ft. (1.8m) weather balloon weights 600 grams and has nominal lift of 3.8 lbs (1.7kg) and maximum of 12 lbs (5.4kg).
10ft. (3.0m) weather balloon weights 1.2kg, nominal lift 7.6 lbs (3.4kg) and maximum 15 lbs (6.8kg).
They want weather balloons to ascent rather fast (several hundred meters per minute), so nominal lift is probably specified for that.
http://www.amazon.com/10ft-dia-Professional-Weather-Balloon/dp/B00513FWQI/ref=pd_sim_sbs_hi_1
http://www.amazon.com/dia-Professional-Weather-Balloon/dp/B004RK2RAU

Considering that 6 ft. balloon is 1.4 cubic meters (50 cubic ft.) and 10 ft. twice that (2.8 m^3) then those blimps that you showed must be very heavy. With 2.6 m^3 of gas they only lift 170 grams and with 3 m^3 a little over 300g.

[RCMR]

For the controller I'd be tempted to use the OpenPilot solution (see http://www.openpilot.org/).

The hardware and software is opensource and it's doing a brilliant job.

Here's a video I posted the other day showing the OpenPilot controller in action:



As for the platform itself, these are a piece of cake to build from scratch.  I'll be posting a "quadcopter build" video in the next week or so, showing the whole design/build process and how the OpenPilot board is installed and configured.

[EEVblog]

Thanks, didn't know about that one, will check it out.
I was going to go with the ArduCopter, as it seems to have the most support out there?

Dave.

[mikeselectricstuff]

For control maybe, but if you are flying this think RC, don't you need a video feed?

Yes, but if it's a slow-moving blimp, you don't need high framerates - a medium-res monochrome feed at, say, 10fps with compression wouldn't need much bandwidth - certainly doable over VHF.