Login to disable ads!

New algorithm makes quadrocopters safer

Sharing your information about aviation, RC, FPV, regulations, any story or even complaints to Parrot global services.

Moderator: AR.Tommy

Jayson Hanes
Spying is fun
Posts: 666
Joined: 23 May 2013, 14:42
Drone Type: AR.Drone 2
Location: Tampa, FL USA
Has thanked: 0
Been thanked: 0

New algorithm makes quadrocopters safer

Post by Jayson Hanes » 04 Dec 2013, 20:09

Yeah.. get right on this, Parrot! Would have saved (just) me at least a few $100 in crosses, shafts, etc.. thus far this year!

http://phys.org/news/2013-12-algorithm- ... safer.html

User avatar
sergeykofanov
Just flips over
Posts: 87
Joined: 23 Mar 2013, 16:52
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by sergeykofanov » 09 Dec 2013, 05:28

Jayson Hanes wrote:Yeah.. get right on this, Parrot! Would have saved (just) me at least a few $100 in crosses, shafts, etc.. thus far this year!

http://phys.org/news/2013-12-algorithm- ... safer.html
"This means that it could be easily implemented on existing multicopters." says Mueller. "The software allows multicopters to remain airborne and perform a controlled emergency landing with as little as three, two, or even just one motor intact."

I can't see how a copter can "remain airborne" with just one propeller or two props at one side. They are liars.

User avatar
sergeykofanov
Just flips over
Posts: 87
Joined: 23 Mar 2013, 16:52
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by sergeykofanov » 09 Dec 2013, 05:35

My big concern is that as the mail corporations become in effect "owners" of the low air, there will be a regulation issued disallowing private drones because of "a chance of collision with corporate drones and becoming a safety hazard to people on the ground", something like that.

pawelsky
Serious flight time
Posts: 1272
Joined: 02 Aug 2012, 22:01
Drone Type: Bebop
Location: Poland
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by pawelsky » 09 Dec 2013, 11:04

sergeykofanov wrote:I can't see how a copter can "remain airborne" with just one propeller or two props at one side. They are liars.
If it can fly with two, why (using the same technique) wouldn't it fly with just one?

http://www.youtube.com/watch?v=w2itwFJCgFQ#t=6m50s

ShellDude
Batterys getting low
Posts: 957
Joined: 04 Jan 2011, 20:10
Drone Type: Bebop
Location: Philadelphia, PA USA
Has thanked: 0
Been thanked: 1 time

Re: New algorithm makes quadrocopters safer

Post by ShellDude » 09 Dec 2013, 16:44

so long as the motor for the single propeller generates sufficient thrust to effect the movement / orientation of the drone then with the right input, calculations, and feedback anything is possible.

great TED video btw!

Ruyton
Made it over nextdoor
Posts: 292
Joined: 06 Jun 2013, 16:39
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by Ruyton » 10 Dec 2013, 09:34

Very clever :D I couldn't see how this could possibly work, but watching the video makes it clear. The drone doesn't keep flying in a stable manner, but does whatever is necessary to achieve 'stability' and maintain height (or a controlled descent) by any means - and if that means spinning on the spot then that's what it does. Yes, I can see this working - certainly with a descent - with only one prop. The drone would be spinning like a top but slightly more controlled than just flipping and crashing :o

I'm guessing that licensing this software to be used in a drone could be expensive?
Gary - Shrewsbury UK

User avatar
AR.Tommy
Serious flight time
Posts: 1328
Joined: 07 Oct 2010, 07:59
Drone Type: Bebop
Has thanked: 73 times
Been thanked: 58 times

Re: New algorithm makes quadrocopters safer

Post by AR.Tommy » 10 Dec 2013, 10:11

Oh dear, this is so awesome! Thank you for sharing.
http://www.digitalsirup.com

Sweet and delicious apps for your drones.

User avatar
sergeykofanov
Just flips over
Posts: 87
Joined: 23 Mar 2013, 16:52
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by sergeykofanov » 10 Dec 2013, 19:10

All the work is done with the sensors inside the spheres and a few RF or IR beams that create a grid to assist in navigation and control. It's a complex installation I believe, and cannot be reproduced with just one or two pairs of transmitters-receivers. The software does its work well, but it really has a lot of data in its disposal.

And re the one prop flight - it's impossible in case of q-copter. You see the demonstration only shows cutting off props on the opposite ends of the line coming right under the center of mass. Any other case won't work. It is necessary that the weight center was located within the area that is drawn around the points of application of the force. In case of only one prop and the center of weight being not directly below it, the system starts rotating around horizontal axis until the weight center positions itself right below the motor. But by this time the prop will be facing horizon and not the sky, leaving no chance for the quad to have anything to propel it in vertical direction and to slow its descent. No hull spinning is going to change the fate.

pawelsky
Serious flight time
Posts: 1272
Joined: 02 Aug 2012, 22:01
Drone Type: Bebop
Location: Poland
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by pawelsky » 10 Dec 2013, 20:02

sergeykofanov wrote:All the work is done with the sensors inside the spheres and a few RF or IR beams that create a grid to assist in navigation and control. It's a complex installation I believe, and cannot be reproduced with just one or two pairs of transmitters-receivers. The software does its work well, but it really has a lot of data in its disposal.
Motion capture helps, but is not necessary (at least for failsafe). Here is the statement from Mark Mueller himself
Q: Does that work outside a Vicon room too? I'd guess a good IMU and magnetometer are enough for a big part of the control right?

A: Mark W. Mueller
The Vicon does make it easier for us, but in principle it will work as long as you have a reasonable state estimate (including some absolute measurements). If you just want a reasonable crash landing if a prop falls off, you could probably get away with only gyros and accelerometers onboard.

The detection is done onboard, using gyros and the motor commands, because this is much faster than the Vicon.
source

sergeykofanov wrote:And re the one prop flight - it's impossible in case of q-copter. You see the demonstration only shows cutting off props on the opposite ends of the line coming right under the center of mass. Any other case won't work. It is necessary that the weight center was located within the area that is drawn around the points of application of the force. In case of only one prop and the center of weight being not directly below it, the system starts rotating around horizontal axis until the weight center positions itself right below the motor. But by this time the prop will be facing horizon and not the sky, leaving no chance for the quad to have anything to propel it in vertical direction and to slow its descent. No hull spinning is going to change the fate.
You don't know how exactly it is done, so claiming they are liars is a bit premature at this moment I think :)

They did it in practice and claim it is possible (and have a patent for that) - let's wait until they publish more details (which they plan for next year)

User avatar
sergeykofanov
Just flips over
Posts: 87
Joined: 23 Mar 2013, 16:52
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by sergeykofanov » 10 Dec 2013, 21:12

I wish I saw how they did it. So far we get no documented evidence. Too often I heard a lot of promises, then once they achieved the marketing goals most promises are forgotten. But more important is that I just can't see how it can physically be done. They need a powerful giro or change the way a motor is attached as to allow it freedom to change the prop angle against the horizontal line. In other words, they must allow the prop keep horizontal while the prop-less part of the drone is falling. But all the extra mechanics come at weight price, which can significantly reduce battery time. There are also complications with controlling of such a system. I doubt it comes cost effective against just adding one extra prop to satisfy the safety.

Jayson Hanes
Spying is fun
Posts: 666
Joined: 23 May 2013, 14:42
Drone Type: AR.Drone 2
Location: Tampa, FL USA
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by Jayson Hanes » 10 Dec 2013, 22:00

:roll:

dragonash
Chased the cat
Posts: 343
Joined: 30 May 2013, 15:57
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by dragonash » 10 Dec 2013, 22:31

wow sergey, very anti saving us all some money, arent we?
Jeez, if they say they have a way or are working on a way to get a safer emergency landing if you were to lose 1, 2 or 3 props, its way better than the emergency landing we have now....which is a dead free fall.

User avatar
sergeykofanov
Just flips over
Posts: 87
Joined: 23 Mar 2013, 16:52
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

New algorithm makes quadrocopters safer

Post by sergeykofanov » 11 Dec 2013, 08:33

dragonash wrote:wow sergey, very anti saving us all some money, arent we?
Jeez, if they say they have a way or are working on a way to get a safer emergency landing if you were to lose 1, 2 or 3 props, its way better than the emergency landing we have now....which is a dead free fall.
Don't get me wrong. I'm not anti saving some money. I wish anyone conducted a poll among quad owners to know how often is it that one (and especially three) props are cut off, to know how much exactly we are to save at the additional cost they are (undoubtedly) going to charge. I also forecast additional costs in terms of reducing battery life, and etc.

The control complications that I mentioned are like if you imagine all the motors are attached freely to the cross ends, say, with a cotton thread, fabric belt or a flexible plastic wire or a joint. Then you won't be able to control quad at all because changing the motor speed will not do anything useful but only send one end of the drone down and another up. Then you need additional hardware to lock the joints stiff while in normal flight. The joins must be unlocked once the system detected the emergency mode. That will let slow down of the descend but render the system uncontrollable. More hardware is required (like locking the drone in the "new" position and controlling by weight shift, like it is in a traditional single rotor heli). We then lose all the advantage of the simplicity of multicopter.

Jayson Hanes
Spying is fun
Posts: 666
Joined: 23 May 2013, 14:42
Drone Type: AR.Drone 2
Location: Tampa, FL USA
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by Jayson Hanes » 11 Dec 2013, 13:54

I'm sorry serg.. everything you've been saying lately makes me shake my head.. it's like you truly have no clue whatsoever what you are talking about.

User avatar
sergeykofanov
Just flips over
Posts: 87
Joined: 23 Mar 2013, 16:52
Drone Type: AR.Drone 2
Has thanked: 0
Been thanked: 0

Re: New algorithm makes quadrocopters safer

Post by sergeykofanov » 11 Dec 2013, 20:57

Jayson Hanes wrote:I'm sorry serg.. everything you've been saying lately makes me shake my head.. it's like you truly have no clue whatsoever what you are talking about.
You said "everything". So far my references were about gravity and points of application of forces that work against gravity. Do you argue that gravity exists? Do you argue that the point where the force is applied is located right below the center of the propeller? Imagine you put the MAIN ROTOR of the traditional single rotor heli not above its center of mass, but at its tail, or at its nose. Will the heli be happy? Will it be able to fly at all?

Post Reply

Return to “General - Parrot / New Products / Info / Story / Complaints”

Who is online

Users browsing this forum: No registered users and 2 guests