Common cause?
"Collision with a cloud of 12-bore shot"
A pilot flying a 3D Robotics Solo drone let his autonomous aircraft fly itself into a crane that some inconsiderate person erected on a building site. The accident, which happened in England in June last year, was revealed in this month's Air Accidents Investigation Branch monthly update. The AAIB investigates aircraft …
Well, it was flying a pre-programmed flight, so it was autonomous, just not self-aware. Time for 3DS to add collision-avoidance cameras like DJI has done? Then again:
https://forum.dji.com/forum.php?mod=viewthread&tid=93915
With a helpful screen-shot of its T&Cs. This flying ace managed to fly his very expensive drone into a power line. DJI was helpful until he threatened them with his advocate.
Nope - this might be the definition of "autonomy" that the drone manufacturere marketing department might like you to use, but true autonomy means having the "freedom to act independently".
A drone constrained to a pre-programmed flight path is - by definition - not free to act independently and hence not autonomous. At all.
Now, if the drone had some additional obstacle detection and avoidance algorithm that would allow it to resumeto a pre-programmed path whilst avoiding unexpected obstacles along the way .... THAT might be considered some degree of autonomy, but since that capability was notably absent in this case then... nope.
I notice that he claims expertise in GPS and that it simply cannot be more accurate than 50meters...
A 50 meter accuracy is so piss poor that it would be unusable for anything on a drone. With the clear acrea in his picture I'd expect better than 10meter, closing on 3.
He was flying it manually, in P-GPS mode, which means he should have had visual control of the drone.
How could he not see the tree he claimed his drone smashed into?
I think he got what he deserves; nothing!
Claims expertise...
Immediately proves he has none.
50m was about it for Joe public original GPS with the locked down military encrypted packet and only a lock on a 3 or 4 satellites. These days you get a lock on close to a dozen with a clear sky like that, and you're down to a couple of meters.
What about the report from the people that put the crane up to the Crane Accidents Investigation Branch? Surely, they must share some of the blame as it's a bit inconsiderate putting a crane up on a building site and doing a quick search it appears that cranes have been involved in a lot of air accidents. Maybe were looking at this the wrong way round.
Since cranes are generally 'up in the air' should they have some kind of licence from the CAA?
I do find it laughable that the operator turned up and flew without doing at least a cursory survey of the site, after all it is a place where 'things' are built, often upwards.
I don't know about now but in my day it took 40 hours minimum to get a PPL maybe they should look into standards for commercial drone pilots, 26 hours doesn't seem much.
I don't know about now but in my day it took 40 hours minimum to get a PPL maybe they should look into standards for commercial drone pilots, 26 hours doesn't seem much.
Prior to this, the last crane versus aircraft accident I recall in the UK was where a highly experienced helicopter pilot with full training and hundreds (if not thousands) of hours experience flew his craft into one, killing himself, in Battersea, IIRC. Anybody wishing the read the details will find them somewhere on the AAIB site.
You can make the training as long as you want, it only tells people what they SHOULD do.
> Since cranes are generally 'up in the air' should they have some kind of licence from the CAA?
If you are near an airport, your crane (or other structure) may interfere with aircraft operations. Your local authority will be aware of this and take it into consideration when issuing the permit. In certain cases, consultation with the CAA will be necessary and they may have the last word. If you are really unlucky and your crane happens to be in the wrong place, it could turn out to be expensive as the safety impact will have to be assessed and a modification of airport operations may be required.
I have no information about the case at hand and no inclination to read the particulars. It is possible that the crane will have been NOTAMed (information about its existence disseminated to aviators), but I have no idea whether drone operators in the UK are currently required to check NOTAMs, or have the training to do so.
Berthed here in Dundee, under the approach path to Dundee airport are three (at last viewing) mobile drilling rigs with their tripod legs sticking very high into the sky. So high they have those flashing aircraft nav lights on them. Mobile drilling rigs are wont to arrive and depart, slowly but in a determined manner.
When they jack them up to the top of the legs so they can paint the bits the hull covers they more than resemble certain SciFi tripod monsters. When it's dark the lights on the superstructures look quite pretty.
Note the Tay is subject to haars, sea fogs which come up the firths and which could hide the tops of those legs to an approaching aircraft.
When crossing the new Queensferry Crossing, the highest bridge towers in Europe, my wife opined she couldn't see nav lights on the towers. I opined the towers were so high we had no chance to see the lights from the bridge deck. BTW they briefly took a connection to the south tower out during construction when it was realised that standing alone with its decking extended the central tower was the tallest cantilevered self supporting structure in the world. They left it like that for long enough to claim the record.
If you visited a construction site with a few cranes on it, then a few weeks/months later visited the same site, would you be able to easily tell if a crane had been added or moved?
Any drone operated autonomously via waypoints needs to have built in collision avoidance IMHO. They should not be allowed to operate autonomously otherwise. It is not just a problem of cranes, what about birds, other drones, hot air ballons, TV helicopters or whatever else might be in the area while the operator stupidly lets it follow waypoints like some sort of drugged up bloodhound?
"If you visited a construction site with a few cranes on it, then a few weeks/months later visited the same site, would you be able to easily tell if a crane had been added or moved?"
If my job was to survey the building site, then yes. Yes I would. Kinda in the ol' job description, no?
If a crane has been moved 30 feet on a 10 acre site it doesn't matter whether that's your job or not you're probably not going to notice it when you step onto the property. That's the whole reason you are using a drone, so you can catch all the changes that are not obvious to your eye and fallible human memory. If you have to memorize the locations of everything, why bother with a drone?
The waters around Benbecula are not "icy" as the North Atlantic Drift comes here and results in wonderfully warm water with huge palm trees along the coast (according to VisitScotland). For a proper Scottish water temperature scale may I recommend: https://www.youtube.com/watch?v=01KVj8kEjMo
Typical operations should involve not only an initial survey to undertake risk assessment as part of the planning process for any mission but also another risk assessment should be conducted on the same day & ideally just before the flight commences to capture any changes in circumstances that had happened between initial survey and now.
Part of a downward trend in capabilities of qualified operators I'm afraid. Anon because I am one.
An AUTONOMOUS drone that was unable to detect and avoid an obstruction might be newsworthy since this would indicate some failing in the claimed autonomous capability of said drone.
A drone that ran into something after being despatched on a pre-programmed flight path is a) NOT autonomous and b) no more than you would expect if the flight path were to contain obstacles that the operator was unaware of for whatever reason.
Move along. No news to read here.
Relying solely on internal guidance with no external control inputs for the duration of the flight seems to fit "autonomous" quite reasonably. (Leaving aside the sensors for height and location, of course).
The lack of ability to make course changes through external threat detection makes it a poor implementation, and can lead to incidents such as the one reported in the article.
If you take the view that there are two states, autonomous or under external control then the drone was clearly autonomous.
If you blindfold someone then turn them loose in a busy street with a few memorised instructions then they wander off under their own control they are no longer under external control. They may be very foolish, but they are making their own decisions and are thus autonomous.
Sense of the term.
IIRC it's pretty much the sense used by the USAF in Vietnam for their drones.
Fly pre-programmed course of altitude, speed and bearing way points, then either execute landing "subroutine" or eject sensor package (the most valuable part) and self destruct.
In 2018, perhaps describing that as "autonomous" is a bit generous.
The drone pilot intends in future to make his drone start by flying straight upward.
If it goes seriously wrong - likely to show up early in the flight - then it will go straight downward. And land on his head.
Okay... I suppose he can stand it on the ground and then retreat a safe distance before starting it.
Not sure the hi-vis would help. The drone can't even see a bloody great crane...
But on a serious note, I'd have thought going straight up is a bad idea. Though the CAA seem to disagree, as they reported that this is what the pilot said he'd do in future, and don't seem to have disagreed.
Helicopter pilots are taught to only hover if they need to, and to always be moving horizontally if possible, as that gives much better options for recovery from an engine failure. Much harder to auto-rotate from hover - plus your own downdraft reduces available lift. Then again the second doesn't apply so much to small drones with electric motors Plus I don't know if they're even capable of auto-rotation - although they do have multiple motors and rotors - so do have some recovery abilities.
Vertical ascend in a helicopter leaves no margin for recovery, vertical decent in a helicopter is dangerous due to the risk of settling with power and/or vortex ring state. For a multi-copter the added efficiency for flying above transition speed also means it's more efficient to have some forward speed during ascend, giving you better flight time and better controllability. The difficulty in seeing what is directly above from the ground and from an inboard camera would also mean it's probably better to have forward speed during ascend. I'm not so sure the chosen remedy here is better than having a pre-made plan of where your obstacles are going to be and then checking on the day if those obstacles are still roughly the same spot. The lack of a pre-flight check here is the problem, not the chose ascend path.
Dear CAA/AAIB. Report 1 (of 15) for 25th December 2017. Inadvertent excedence of upper altitude limit.
At approximately 10:05 am on the morning of the 25th a drone operator took off in near proximity to several bystanders. Unfamiliarity with the controls resulted in the craft ascending rapidly. At an altitude of approximately 2.3 metres the drone reached a hard altitude limit (also described as a ceiling). The impact with the ceiling detached a propeller, and the drone operator cut power. An uncommanded descent resulted in impact with a bowl of twiglets, but no further damage to the drone and no injury to bystanders. The operator conceded that a) he shouldn't have climbed so high, b) he shouldn't have stolen the drone control from his nephew and c) he shouldn't have flown after downing two glasses of bucks fizz in 10 minutes.