On predicting in flight collisions

[kgwilson]

Agree in the circuit ADSB may as well be turned off. It would just be a distraction. On XC with up to 40NM range it is very useful. Gives you time to change course if you think there is any conflict & you can always call the other aircraft as the rego is displayed, but is he on the same frequency as you?

[RFguy]

I dont agree about ADSB being turned off in circuit.

circuits can be busy. and radio comms can be crap. 

but it needs to be in a a mode that adds something to the already busy pilot. 

[kgwilson]

Wrong choice of words. I never turn mine off. I just don't look at it in the circuit as I am looking at all the traffic and making my calls.

[Markdun]

It all depends. My main motivator in getting and using the SkyEcho was to avoid heavy military Rotary aircraft at circuit altitude & below over my airstrip. I missed one by several hundred metres on downwind on one day; they didn’t respond to my radio call and subsequent correspondence with the military was that they just use ‘see & avoid’ in class G.  So yes, I currently use ADSB for in circuit at my place.... but at Goulburn, no, because most aircraft don’t have it turned on or installed (you would think training aircraft would have ADSB out wouldn’t you?).

[Garfly]

Yeah, in some ways, the circuit is the most useful place for it. Obviously, your eyes will be 95% outside and on a swivel, but if it's well set up and used, a traffic screen needn't be more distracting than your ASI, say, or your downwind checks.  Sure, less than half the VFR fleet is currently on-board but that's changing fast. Anyway, by the time you're near the airport, you might have sussed out which of your chatty circuit mates are true icons and which not.  At least all commercial machines in the mix should light up.  Might save the travelling public, at least. 

I've breathed quite a few sighs of relief near airports, catching possible conflicts on the fish-finder.

For sure, my best efforts at converting circuit-chat to geo-locations are never as good as the wee pics on the iPad (and never as negligible a distraction).

 

 

 

 

Edited November 14, 2022 by Garfly

[Carbon Canary]

The training area for Camden + Bankstown in Sydney currently sits over the top of the rapidly developing new Badgerys Creek international airport.  Its listed as a danger area for a good reason.  Its congested with lots of student traffic doing what students need to do.

 

In this area ADS-B with aural alerts and relative position data has identified a heck of a lot of traffic.  Many alerts may be regarded as a nuisance, but on occasions, potential conflict has been real.  The circuits at both Camden and Bankstown are managed by ATC, so no problem there.   Many, many years ago there was a fatal mid-air over a VFR approach point for Bankstown which I believe was the catalyst for creating Class D.  To this day however, the VFR approach points are possibly still the riskiest location within what is already a danger area. 

 

The (rental) aircraft I routinely fly is also fitted with FLARM, but to date I haven't experienced any FLARM alerts, possibly because the gliders at Camden are well separated from power traffic.  I have heard other pilots mention they have had FLARM alerts around Temora.

 

It was this experience that encouraged me to order both ADS-B in/out and FLARM for my own aircraft.  I can tolerate the occasional nuisance alert rather than never getting the critical one.  As mentioned previously in another thread, it maybe out in the boonies where two aircraft are flying the same or reciprocal magenta line that ADS-B may make the difference to being around to fly another day.  

[sfGnome]

Coming back from Ausfly one year, I was chatting with my passenger about nothing in particular, when another aircraft crossed our path diagonally from the rear, maybe 100ft above us. Regardless of swivelling eyes, we never could have seen it due to it’s location, so our continued existence relied on its pilots being aware of us. 
 

Out in the boonies, yes, but ADSB would have been really helpful… 😳 In that case, it wouldn’t have even needed a particularly sophisticated collision detection algorithm, but an aural warning of the location would have been great. 

[Markdun]

Yes. ‘See and avoid’ is total bullshit.  The only planes you see are the one moving with respect to your aircraft, ie. these are the ones that are not on collision course. In other words you automatically will avoid the planes you see., so in that sense the planes you see you will avoid. The ones you don’t see, i.e. the ones not moving in relation to you, are the ones you won’t avoid, but need to.  

On the boat the only way you tell if you are not on a collision course with another moving boat or a stationary object like a rock is to take a bearing; if the bearing remains constant over time you’re going to crash. It still requires you to see the non-moving (with respect your vehicle) object in the first place and have time to assess it.  But in the air, IMV, ‘see & avoid’ is simply relying on the low probability of a collision as others have said earlier.

[facthunter]

Way back when I started I used to go where others didn't and get low to see them. I can't recall any near misses during that time .When everone tracks along the same route near a busy airport you are setting yourself up. Big sky is in your imagination. Generally there is fairly limited level choice below 010. !0 000. En route also people continue to pop out of nowhere and a near prox is close to a disaster. The " a miss is as good as a mile" is not very valid, Vertical separation is safest even if it's you working with another plane. Radio services have been reduced for planes flying OCTA for a long time now.to save $$$'s Perhaps they  should be reinstated in specific well trafficked areas and IF you radio is scratchy etc you go somewhere else. Radio procedures are at I would say an  all-time low. How many could accurately say after a request for your position where you are in relation to an aerodrome. IF you give an erroneous position the situation is worse than if you didn't reply at all.. Nev

[Garfly]

On 15/11/2022 at 11:14 AM, facthunter said:

 How many could accurately say after a request for your position where you are in relation to an aerodrome. 

Actually, just about everybody nowadays. A single tap on your EFB/tablet, shows your NM/degrees FROM a list of every airfield within cooee. As well as any other type of aeronautical waypoint (editable for taste). 

[Markdun]

 Most calls aren’t ‘5nm SE, 4500’, they are ‘over the chook farm’, or ‘training area, 4500’, which is just dandy if your are unfamiliar with the airfield.  Again in boating GPS derived Lat,Long in degrees & minutes has de facto become the standard for position reporting; no more 15nm NE of South Solitary. Only problem is many boaters (& coastal ‘marine rescue’) seem to think giving minutes of Lat & Lon to 4 decimal places is necessary.

 Are we drifting from the topic? I’d like to know if RFguy has written the code yet for a software upgrade for Dynon, MGL EFISes with ADS-B in or OzRunways etc that will provide proximity/safe separation info and warnings?

[facthunter]

Predicting must contain factors involved discussion not just gadgets as a  solution. Any person flying in the vicinity af an aerodrome should be already aware of where they are relative to it and ON the appropriate frequency to talk to others in that area regardless if how that is determined, Nev

[Roundsounds]

On 13/11/2022 at 11:50 AM, RFguy said:

Hi Bruce

Most seem to be circuit issues, but a few have happened en route. 

 

My $64 question for the audience is "what should it do when it detects a collision is likely" ?

 

I did a heap of research into midair collisions in Australian airspace, couldn’t find any enroute aside from people flying in company / formation. Most collisions occurred in CTA. 

[Ian]

Out of curiosity what language and library have you written this in and how computationally expensive is it?

 

[RFguy]

Hi Ian

It's currently written in a mix of python and C++. It will need to be all C++ to go into an embedded platform.  

Computationally- fairly intensive as there are alot of matricies to solve regularly, but they're only 4x4s mostly.  I'll figure out the computational requirements  down the track (no pun intended) 

Fortunately it only has to calculate everything every second for each update that arrives. And it can sensibly load shed basied on separation if it needs to.

 

I'm not sure right now what the embodiment will be. I am thinking the standalone box with outputs for warning indicators, intercom audio , a large say 24x1 display.  

Ive also experimented with a 3d looking 2D display plotting points on the surface of a sphere that has your airplane in the centre. .

 

It's possible it could be integrated by 3rd parties into existing applications on tablets. those things have plenty of transient processor power available. 

 

I'll do a bit more on this in December when I get a bit more time . 

MarkD- it could be a box with just inducator out, audio out, RF in ---that also has an output  to talk to a MGL, EFIS etc. IE the black box does the computational hard work, leaving another box to display

I presume its possible to put  messages onto an MGL screen over everything else /.

 

 

 

Edited November 17, 2022 by RFguy

[Bruce Tuncks]

We need to have an automatic system ( eg the flarm ) which consists of a transmitter/receiver and a computer and ( I think ) a gps. All it does is warn the pilot and that is more than enough I reckon. They sure keep you looking out the window.

En-route collisions are more likely since GPS systems came into use. They provide a concentrating factor, but with a bit of software, this can be overcome.

[Bruce Tuncks]

Sorry Glen, we were posting at the same time. I like your idea, and if the price is only a few hundred dollars, then it is just what we need.

[Ian]

20 minutes ago, RFguy said:

It's currently written in a mix of python and C++. It will need to be all C++ to go into an embedded platform.  

You may be able to leverage an openCL Kalman filter on an embedded platform using an onboard GPU. I've got a multicore embedded AMD Ryzen board somewhere with fairly powerful integrated GPU somewhere that would probably be overkill but OpenCL provides an abstraction between CPU and GPU which is good for this type of operation

[RFguy]

Hi Ian

I'll be looking  at a processor that uses  less than 1W. Most likely a medium 100,000 logic cell   FPGA with soft processor, using FPGA fabric as hard matrix solvers. Done that before- familiar territory for me....

Edited November 17, 2022 by RFguy

[Garfly]

On 15/11/2022 at 10:26 AM, sfGnome said:

Out in the boonies, yes, but ADSB would have been really helpful… 😳 In that case, it wouldn’t have even needed a particularly sophisticated collision detection algorithm, 

This Canadian trip-video (04:30 to 09:00) includes a good illustration of how simply useful a simple ADSB system can be, VFR enroute, even when radio comms are established. (Especially if hemispherical level-keeping constrains vertical separation - as might be happening here).  They were probably miles apart at all times but you can see how much anxiety was caused.   And how easy the solution can be.

 

 

Edited December 23, 2022 by Garfly

[kgwilson]

Instead of trying to figure out where everyone was at the same altitude I'd have changed to the next VFR cruising altitude rather than doing a couple of 180s given that one of the aircraft was behind him on the left or was it the right, he wasn't sure. That would have reduced the stress level and got them there quicker without much difference in fuel burn. Turning around to fly towards an aircraft somewhere behind you at the same altitude is not smart.

 

If all 3 had ADSB then the use of that and the radio for confirmation of location would have reduced stress dramatically.

 

Interesting terminology. A loop in my book is a vertical 360. His were 2 x 180s in opposite directions. Also height, forty five hundred instead of four thousand five hundred. When he originally said loop I assumed that he meant an orbit.

[Mike Gearon]

On 15/11/2022 at 11:14 AM, facthunter said:

Radio procedures are at I would say an  all-time low. How many could accurately say after a request for your position where you are in relation to an aerodrome. IF you give an erroneous position the situation is worse than if you didn't reply at all.. Nev

We had a ripper recently. Cloud was lowering  departing aircraft ability to fly out above circuit height. We requested a departing aircrafts position as we started downwind entry and told it was “Near the ship”

 

It wasn’t great. That aircraft was departing at our downwind entry point and height and we spotted each other up close and personal.

 

 

[pmccarthy]

The Skyecho has made my relaxed and enjoyable flying a lot more stressful, looking for traffic I previously knew nothing about!

[Bruce Tuncks]

And there was no midair possibility for most of those other planes. That's why we need an automatic computer system to process the information that is causing stress right now....  I hope you are going good there rfguy.

One problem the flarm had...  it stopped working ( I think yearly ) until you paid them more money, and the other aircraft needed one too. The money was not the problem, it was the way it just stopped working.  

[Tasmag]

Sratux is a good option for the MGL iEfis, talks to it via RS232 and gives audio warnings based on height and distance you set.

 

Total cost of about $120. (possibly a little more post COVID, I haven't checked recently)

 

Mike