Quads landing 'safely' after prop loss

[quadfather]

Have a looksee here:  http://goo.gl/1mtl86

Researchers at ETH Zurich have developed a novel control algorithm that allows quadrocopters to continue to fly in spite of multiple motor or propeller failures. This algorithm makes these vehicles safer and may allow them to be used for instance in delivery services.

Cheers,

Tim

[GdLookinMingeR]

Will be interesting to see if the code roles out
as a general release or sold only to closed systems such as dji.

[n506]

Will rely on some kind of code to figure out what way the quad is facing at all times, because otherwise it';s just a method of slowing down the descent. That could be straight into a river, onto a house, or onto a person. The ideal is that it instantly heads for a known safe location (RTH style) using GPS while calculating its rotation speed so that it constantly adjusts in order to move in the required direction.

[Biffa]

I can';t honestly believe that people really think that any kind of multi rotor would be used for day to day deliveries.

[Toledodave]

Biffa I agree with you. These things are not stable enough to turn your head and walk away. Not forgetting to mention the liability for the ppl at the recieving end in the event of injury.  With all that being said according to amazon they are serious about putting this in action in the USA. They claim they are only waiting for FAA to create and adopt new laws and rules to begin implimenting system. Also I thought I read somewhere they wouldnt be using gps as if a different means to triangulate coords. But lets face it..as long as there is investors there will be research and development. Maybe...just maybe they xould do it but I believe it wont be done with the technology at hand now. Just my opinion.

[GdLookinMingeR]

Will rely on some kind of code to figure out what way the quad is facing at all times, because otherwise it';s just a method of slowing down the descent. That could be straight into a river, onto a house, or onto a person. The ideal is that it instantly heads for a known safe location (RTH style) using GPS while calculating its rotation speed so that it constantly adjusts in order to move in the required direction.

I think your completely missing the point this code recovers from prop loss and recovers, it calculates a required algorithm to maintain the copter in flight at that point you could ask it to RTL.   DJI have been doing this with the hex and up config but to actually recover from a quad loosing a prop is a amazing bit of work. 

DJI F550 prop loss mid flight from Hexa to Quad

[Hands0n]

This kind of control was demonstrated on a TED talk earlier in the year.  It is an amazing bit of control recovery, and as a FailSafe it could be quite essential.  Although looking very untidy, the FailSafe flight was in stark contrast to the demonstration of the aircraft with it disabled.  I would love to know that in the event of such a failure my aircraft would stand a chance of being recovered instead of turning into a stone and responding to Newtons law.

[quadfather]

...instead of turning into a stone and responding to Newtons law.

Don';t forget the three rotating hatchets! 

[Biffa]

The TED thing was "amaze balls" to coin a current phrase! And would be awesome if implemented to the hobby side of the tech

To even think the FAA would even sanction such stupidity is preposterous, especially as they are giving the TBS guy so much grief. Having said that, if all of this hype somehow contributes to us at grassroots level then that';s all good

[Hands0n]

I';d tend to agree. At this precise time in the evolution of these aircraft I can';t see how they can be safely used for autonomous transportation. But given sufficient time to design in the essential safety features it may well come about. Basically, I';m not saying "No", but have a healthy scepticism about the whole thing.

But yes, such developments and stories are only going to do the hobby and the aircraft some good. They have been demonised by the mainstream media until lately. Just as long as the government';s "forces" do not take our toys away for their own exclusive use I';ll be happy.

[n506]

I think your completely missing the point this code recovers from prop loss and recovers, it calculates a required algorithm to maintain the copter in flight at that point you could ask it to RTL.   DJI have been doing this with the hex and up config but to actually recover from a quad loosing a prop is a amazing bit of work. 

DJI F550 prop loss mid flight from Hexa to Quad

It doesn';t "recover" from prop loss... it completely alters all flight characteristics in order to stay in the air. You have a constantly rotating situation meaning that flying in a straight line in any direction is an immensely complex situation of having to add additional input to all three remaining props not just to stay upright (which it clearly can';t quite do from the video so far), but to actually continuously adjust to move in a given direction.

Add to the fact that that heading direction is spinning round in circles. Try walking down a road spinning as fast as you can and see how easy you find it. I';m not even sure that a compass magnetometer would be that accurate at such continously high spin rates, let alone the confusion to any gps systems. Add to that the fact that a prop almost never comes completely off, but instead will break on one side, resulting in insane amounts of vibration, and I think this was what you';d call a very easy test for this new software.

For it to be viable, it needs to be able to cope with immense vibration and be able to understand its position and direction through all of the spinning. This is very easy for a hex or an octo to do in the event of a lost prop (albeit vibrations can still be a huge issue), because they don';t have to spin to stay upright thereby adding such huge complexity to the situation.

When they can overcome a partially damaged prop and still maintain heading in any direction, I';ll be impressed. That said, FPV would be fun!

[kilby]

The safeish landing  is a good thing, though the max yaw parameter may cause some issues

As for the Amazon delivery business the stupid things like where the packages are left and the completely trivial range issues (matters tech is not advancing that fast) I mean things are bad enough with the so-called couriers that have no concept of numbers and post office employees who think people sleep next to their front doors, just in case something needs to be delivered

If it happens then its going to be a very short range gimmick

[Toledodave]

Dont forget even if we accomplish better communications and gps function we still have a major issue...."payload". We are striving with multis to get as long as flight times we can but you still need to balance flight times vs payloads. We all know you cant have both. So from what ive seen and tryed my Y6 weighs in at 7 lbs and roughly 9 min flights with no payload. I believe I can carry another 5 lbs but for how long. I still believe we are long way from commercial multicopter flights. Unless we advance past the Lipo.

[n506]

I';m not so sure it';s a no-go even with relatively current technology... I could foresee a "London" 30 min delivery service. Perhaps, for example, a fairly manageable number of staff positioned at strategic points. They receive half hour deliveries from one or more quads from the central warehouse, which they then deliver within say 5 to 10 blocks distance before returning back for more. None of the hassles of traffic issues across the city, because they could use bicycles for such small delivery zones, and delivery would effectively cost the price of wages and lipo charges for a relative few bucks worth of quads. Take out the cost of vans, fuel and also factor in the lower emissions and you have a very practical solution for city areas within the 10 mile radius of the city centre. The inner city drop-off points could also act as recharge points for deliveries further afield potentially too.

For anything heavier, they';d just outsource to traditional couriers, but I';d imagine the vast majority of amazon sales would be well under the weights a quad could happily lift.

I do think the reality of people getting deliveries by a quad directly is unlikely any time soon. That said, you could potentially see amazon paying people a small rental to host a stop and recharge point which would have a conveyer belt system with say 10-20 lipo packs up on a pole. The quads would land, drop off a pack onto the conveyer, and the conveyer would immediately start charging that pack, while rotating a fully charged and balanced pack into position to be picked up by the quad. This operation could take less than 10 seconds before the quad was on its way further afield. The routing software would plan a route based on wind conditions measured at each of these locations, and how many stops were necessary for the conditions to make it to the desired final stop.

Service quads would also come and pick up any batteries which failed automatic testing, and return them to base for servicing or disposal.

[Hands0n]

While Amazon';s announcement is big and bold there is an element of reality in the suggestion that we could use multi-rotor aircraft for deliveries.  A recent TED talk was very inspiring, considering what the baseline suggestion was and the intentions of the "project".  Well worth a watch for ideas alone.

http://on.ted.com/Matternet

So the thinking is that the power problem could be eased, if not solved, by having appropriately positioned way stations that where the autonomous aircraft could swap out its power source for a freshly charged unit and then continue on its journey.  That would be completely do-able by todays technology.

All it takes is the will and some cash to make it a reality.

[Hands0n]

Back on topic - go to 6:46 into this video from TED

http://www.ted.com/talks/raffaello_d_andrea_the_astounding_athletic_power_of_quadcopters.html

The rest of the video is worth watching also ::) But hopefully what you';ll see will indicate what can be done, not just land, but safe(ish) flight.

[Toledodave]

I just wonder LOS with GPS in citys with tall buildings. As soon as a few signals drop due to buildings the UAV cant precisely allocate its waypoints therefore raising odds of failure. Out in the country it would have better chances. Just my opinion :/

[flybywire]

Well, I guess a couple of decades back, McDonalds'; drive-thru';s were science fiction (here at any rate), so anything';s possible.


Also driverless cars have got green light now.

[Hands0n]

I think that Mobile Phone technology has shown us the way through the limitations of GPS.  Assisted GPS or a variation on it will solve all of the positioning issues - with millimetre accuracy no doubt.  The technology exists already, even if it is out of reach of "civvy street" right now.  Only a few years back GPS with the accuracy we now take for granted with our multi-rotors would have been illegal.  It isn';t the technology so much that is holding this stuff back.  Its mostly legal and chemistry (battery tech).

[kilby]

I think you mean Differential GPS which iusually used by the guys erecting stuff out at sea to keep the cranes in the correct position. The kit i saw demonstrated used multiple GPS receivers and  GPS beacons

Assisted GPS usually asks a mobile phone company for the location based on the mobile phone towers it can find and relative signal strength. This is often used to provide an approximate location while older phones searched for the GPS satellites and started working things out.

Google still use it (along with WiFi ssid beacons) location services but it';s only accurate to a few hundred metres  (without using regular GPS)

Also covered phones that had no GPS chip set, but the company could still advertise A-GPS

It also let them comply with the US laws that demand that the location of all phones in the US could be located

[kilby]

BTW if I sounded snotty it wasn';t intended but I';m currently being forced to use a 50 quid android tablet and it';s fairly hard to post something before it falls off the WiFi network (wife murdered the macbook PSU so she is planning the Christmas shopping on the other machine)

[Hands0n]

Actually, snotty  , I did mean A-GPS - but only as an example of what could be done in urbanised areas by use of a implementation of suitable technology. My bad, I didn';t explain myself properly.   

So what I';m getting at, is that whatever GPS problems are currently insurmountable I do believe that such autonomous delivery vehicles will require an infrastructure that is more than what we see today. But possibly leveraging on new ideas and even the incumbent technologies in a suitable mix. 

I am loathe to say never, given that there is much technology that never sees the light of day - not trying to sound like a conspiracy theorist. But what we have witnessed in the liberalisation of GPS technology over the past few years, unthinkable perhaps a decade or two ago, leaves me in no doubt that we have plenty more to come.

It does feel that, with autonomous multi-rotor vehicles, we are just a little way down the Wright Brothers road in terms of safety and practical usefulness.

[kilby]

The gps a to b business to me wasnt a huge thing against it happening though suitably high quality mapping to make enhanced GPS useful is a completely different (and much more expensive) matter

Power and telephone lines are not easily avoided and sonar or computer vision isn';t up to it, anything short of mm wave radar. We still have a lot of power lines marked for visibility (racing pigeons still hit them)

Perhaps servicing outlying pickup points from local distribution warehouses with some high paying subscribers having to the for deliveries (after paying for a suitable survey)

After all Amazon won';t want to lose a UAV because somebody moved the washing line or let the kids put a badminton net up

Although it would be great to see such things happening it doesn';t make up for the jetpack that the future promised

[Hands0n]

Well I dunno about you, but I';m still holding out for the Jetpack 

[Toledodave]

Lol jetpacks.  I dont think the gps issue is an issue. There are lots of options not forgetting to mention the financial investor support for the technology. Biggest factor I see is just power. Lipos arent going to cut it. Sure they will manage a short delivery but the recharge time wont make it efficient. Especially with payloads varying and forcing unconsistant flight times with liability threats they will have no control of.  They will need like 30 min flight times. 15 mins in and 15 mins back to home to be able to cover a respectable delivery area. Also keep in mind... more then likely they will maintain max alt limits untill destination and drop straight down to landing zone in orser to avoid most city obstacles like utility lines etc...    with all that being said I also believe the government has better power sources then a lipo. If thwy are willing to share that technology then we have something. It wouldnt cost much to adapt and absorb all the gps and cell towers and even add their own triangulating system if need be. I cant totally rule out the possibility of amazon making it happen but they wont be able to with current avail technology is all. In my opinion.

[Hands0n]

I tend to agree with you there.  With precisely the incumbent technology this is a massive ask.  But with the right development there may be opportunity leverage much of what we already have, even cranky LiPo with sufficient "battery farms" from which the vehicles can pick up a freshly charged unit.  But masses of logistics to solve, not the least being the dreaded Elf n Safety.

[quadfather]

I wrote to them asking if the algorithm will be released to the public domain so that open source developers can benefit from it.

The response:

Thank you for your interest. We have applied for patent protection for the system, and are interested in commercial aspects. The algorithm itself is, however, described in a publication, which we expect to be published in June of next year.

Doesn';t sound like OSS to me. 

[Friskle]

everyone will have an opinion on this, mine, is that if i cant fly my quad near buildings or people, then why can someone else, just because its a business.

considering so many rules regarding RC aircraft, if this idea is allowed, it really puts dam near every rule out the window, imho, this is just asking for trouble, and possibly in the long run, make the hobby flyers life harder work than it already is.

[Toledodave]

Friskle,  I agree for most part. I would have to believe the FAA will have rules and regulations reguarding some choices which will be seperated by limitations and money and different classifications. Therefore hobbiests will have freedom under certain altitudes and zones to fly for free. Now the advanced pilots for a fee will be able to aquire a license and show like a proof of insurance to protect the public cover more zones and higher alts. Then I bet when you go commercially it will require even more licensing and more money since comercially its for making money. Therefore the government will be able to charge a nice fee for the commercial license and grant more privilages for commercial use just like any other buisness. So like I said I agree but.... :/

[Jumpy07]

Multis with 6 or more rotors.. with DJI flight controllers.. will happily fly with a loss of one rotor.. other FC';s do similar things..


On DJI FC';s they will start to spin depending on the payload, they do this to retain stability...   you have to recognize when this happens and turn on course lock mode so you can control direction and land safely..


I have reservations if a quad could do the same safely.. the laws of physics tell me it would have a tough time.. it would need to spin fast to keep level I think..   its all going to depend on the weight and payload..