On predicting in flight collisions

[turboplanner]

44 minutes ago, RFguy said:

OME

yes indeed. you'd expect 55-100 kts in circuit for most GA.  ... and  you only get one KNOWN update per second (although you can generate predictive data at any rate) 

 

There's alot that can be nutted out . IE time based  / conflict likelihood based etc hybrid. I'll have to read up a bit on TCAS.

You also have the 1500 and 500' circuits at many airports 500' will be under 65 kts, some by a big margin, but overtaking the 1000' aircraft to the landing zone becaue of their shorter circuits. The 1500' entry aircraft will be travelling quite a bit faster, sometimes on the same route. Helicopters on the Bass Strait rigs will be in the circuit mix, and that's another lot of speeds from R22 to multi passenger Bass Strait gas well, Hay Point ship's pilot and other commercial operations where they don't want to mess around, and you have 2 or 3 12 aircraft per circuit double circuit airports, but mostly there are tower controllers helping separation there.

 

Edited November 13, 2022 by turboplanner

[facthunter]

Realistically it might come to having the equivalent of the Aeradio in congested areas and mandating a serviceable radio. Check it before you fly.  Nev

[Thruster88]

1 hour ago, RFguy said:

 

 

"Conflict   3 o'clock high distance 400 meters Traffic  kilo hotel alpha" is a good start. If thruster88 is coming in from 3 oclock high 400meters away, by the time you comprehend this, he's probably 200m away. 

 

I would love to have some sort of 3D 'ball' on the instrument panel that you could see the planes in a sphere around you like little bright dots.

 

 

 

If the conflicting traffic is 5000m away by all means give the aircraft call sign. 2000m or less replace the call sign with a recommendation,   Conflict  3 o'clock high distance 400 meters, recommend descending left turn. 

We can all work out the solution instantly whilst siting at a desk, flying there will be the startle factor, passenger talking on the intercom etc.

[RFguy]

OME - essentially- how it works - 

1) Acquire time series location data  (from ADSB receiver from other aircraft)

2) apply a Kalman filter to that data to get the most likely values- since the raw location data from the ADSB probably has a random magnitude of 5m in XY and 10m in Z.

3) Now we have cleaned data, apply the data to a non linear regression  prediction tool . There are many- and quite sophisticated tools and algorthms. 

4) predictions are done, paths drawn in 3d space, and  for each combination of airplanes, distances calculated as a time series to provide a probability of a conflict

 

Think of the use of a Kalman filter as a zero lag  averaging tool. 

[facthunter]

Maybe I'm reading it wrong but don't go near advising another plane what to do directly. Look above and to your left would be ok (example) Just imagine IF you CAUSED a death with the best of intentions . You are not licenced to issue instructions. Having said all that I have complied with a turn left immediately  from an unauthorised source in relation to a descending chute which was going to land on the active runway.. Nev

[RFguy]

Hi Nev

no, wouldnt be advising another plane. I'm still in favour of not providing any sugegstions, just information, but Thruster is right that when it gets close, you need somethign like "execute decending left turn"...  if the prediction model is doing its job, this info will come fairly early.... the problems come when say you are goinging on base (why?) and someone has not yet turned base, but is late downwind.  the predictor thinks (without extra information which it could learn....)  that you are continuing downwind. Only when you start the turn that the predictal will go ah oh when it plots your new likely trajectory. IE once the yaw rate ... 

 

The predictor is not specifically working from XYZ, its working from calculated yaw, pitch, roll rates (since they are good ways to describe aircraft motion)  and velocity that is guesses from the XYZ data. 

 

 

 

 

[Garfly]

2 hours ago, RFguy said:

"Conflict   3 o'clock high distance 400 meters Traffic  kilo hotel alpha" is a good start. If thruster88 is coming in from 3 oclock high 400meters away, by the time you comprehend this, he's probably 200m away. 

In his RV, maybe, but in his Thruster? nah, he'd barely have moved (other than to 6 o'clock  ;- )

 

[Yenn]

How does flying the hemisphericals make it more dangerous?

[facthunter]

By giving you vertical separation I don't think it does but you do need to do more Like how you treat waypoints. Tracking and making calls. Nev

[Flightrite]

I don’t know how I/we ever survived 40-50 years ago when the skies were very busy and GA was booming! Too many gadgets these days for VFR drivers getting distracted! KISS still works!

[facthunter]

What , The ground after landing?? Nev

[jackc]

Well, I thought of a goofy idea today?  How about ADS-B data includes the aircraft’s current radio frequency so that data is visible to all nearby aircraft.  People’s ADS-B broadcasts  could alert said aircraft of the frequency that are on and IF the receiving aircraft could be alerted IF they are on a wrong CTAF or Area frequency?  Also IF you wish to speak to any aircraft nearby, you know what frequency they are on? 

[Thruster88]

6 minutes ago, jackc said:

Well, I thought of a goofy idea today?  How about ADS-B data includes the aircraft’s current radio frequency so that data is visible to all nearby aircraft.  People’s ADS-B broadcasts  could alert said aircraft of the frequency that are on and IF the receiving aircraft could be alerted IF they are on a wrong CTAF or Area frequency?  Also IF you wish to speak to any aircraft nearby, you know what frequency they are on? 

If you have the traffic information then adjust your flight path to avoid the conflict. When driving and we don't talk to the other driver if conflict is imminent.  

 

As an avid watcher of dashcam some drivers do try to talk to the other driver, it never works.

Edited November 13, 2022 by Thruster88

[APenNameAndThatA]

Tempted as I am, since FR Guy actually did the the maths on a kalman filter, I will not attempt to explain to him how they work. 😆

[APenNameAndThatA]

Tempted as I am, since FR Guy actually did the the maths on a kalman filter, I will not attempt to explain to him how they work. 😆

 

Also, since aircraft turn it probably is a good idea to take acceleration into account. 😆

 

Maybe proximity detectors use this maths, maybe they don’t, maybe the manufacturers  would refuse to tell you.

 

The great thing about this system is that it would work around airports: there are plenty of aircraft very close that do not pose a collision risk. 

[APenNameAndThatA]

Tempted as I am, since FR Guy actually did the the maths on a kalman filter, I will not attempt to explain to him how they work. 😆

 

Also, since aircraft turn it probably is a good idea to take acceleration into account. 😆

 

Maybe proximity detectors use this maths, maybe they don’t, maybe the manufacturers  would refuse to tell you.

 

The great thing about this system is that it would work around airports: there are plenty of aircraft very close that do not pose a collision risk, and they would not trigger an alert. 

[APenNameAndThatA]

Tempted as I am, since FR Guy actually did the the maths on a kalman filter, I will not attempt to explain to him how they work. 😆

 

Also, since aircraft turn it probably is a good idea to take acceleration into account. 😆

 

Maybe proximity detectors use this maths, maybe they don’t, maybe the manufacturers  would refuse to tell you.

 

The great thing about this system is that it would work around airports: there are plenty of aircraft very close that do not pose a collision risk, and they would not trigger an alert. 

[old man emu]

2 hours ago, APenNameAndThatA said:

since aircraft turn it probably is a good idea to take acceleration into account.

I'm concerned about a little bit of semantics here.

 

Are you using "acceleration" in the sense it means in physics - rate of change of velocity - and saying that a change in the direction or magnitude of the velocity vector is acceleration?

 

OR

Are you using "acceleration" in layman's tems of simply an increase in "speed"?

 

I have been thinking of that word in layman's terms as aircraft don't usually accelerate ( speed up or slow down) rapidly enough to greatly affect the received velocity data. That's why I say that determining acceleration would add extra steps in the computations that might not provide any better results than if it was left out.

[RFguy]

There's acceleration  as a vector, and that is allows you to make a turn of course. no acceleration = straight line flight.

[old man emu]

1 hour ago, RFguy said:

There's acceleration  as a vector, and that is allows you to make a turn of course. no acceleration = straight line flight.

I don't think you are getting my drift about which meaning of "acceleration" you are using.

 

I fully agree that when the word is used in Physics, an alteration in the direction of a velocity vector results in a change in "velocity" because velocity in Physics involves rate of movement and direction of movement. Change either over a period of time and you have acceleration. 

 

On the other hand, the non-Physics use of the word simply means to increase or decrease speed which is rate of movement without direction defined.

 

In real life, I would expect that incoming data relating to another aircraft IN MOST SITUATIONS would show a constant velocity(speed) of that aircraft. Therefore, to require the software to calculate accleration from the data would be an unnecessary task in light of the overall goal of the software. Completely different requirment for the software if it was being used to identify and predict the intercet point of an incoming missile.

[Neil_S]

14 hours ago, Yenn said:

How does flying the hemisphericals make it more dangerous?

 Example : I am flying a course of 045 degrees, at 3500 ft, someone else is flying a course of 135 degrees also at 3500 ft. Both of us are complying with the hemispherical rule, but we are potentially heading towards each other on a collision course. 

 

I would recommend you review the extensive work done by Robert Patlovany by googling "Robert Patlovany collision", and also the work done by Russell A. Paielli. 

The 1928 Australian invention, the Altimeter-Compass Cruising Altitude Rule (ACCAR), would be much safer. 

 

Paielli's model, made in 2000, corroborated an earlier 1997 model by Patlovany showing that zero altitude error by pilots obeying the hemispherical cruising altitude rules resulted in six times more mid-air collisions than random cruising altitude. Similarly, Patlovany's computer model test of the Altimeter-Compass Cruising Altitude Rule (ACCAR) with zero piloting altitude error (a linear cruising altitude rule similar to the one recommended by Paielli), resulted in about 60% of the mid-air collisions counted from random altitude non compliance, or 10 times fewer collisions than the internationally accepted hemispherical cruising altitude rules. In other words, Patlovany's ACCAR alternative and Paielli's linear cruising altitude rule would reduce cruising midair collisions between 10 and 33 times, compared to the currently recognized, and internationally required, hemispherical cruising altitude rules, which institutionalize the navigation paradox on a worldwide basis.

The ACCAR alternative to the hemispherical cruising altitude rules, if adopted in 1997, could have eliminated the navigation paradox at all altitudes, and could have saved 342 lives in over 30 midair collisions (up to November 2006) since Patlovany's risk analysis proves that the current regulations increase the risk of a midair collision in direct proportion to pilot compliance.

Edited November 13, 2022 by Neil_S

[kgwilson]

You are on a merging course and yes a collision could occur. Both at the same altitude and without ADSB what about lookout/situational awareness? Don't just look ahead or bury your head in the cockpit looking at the Nav screen, scan the sky all the time. It helps keep you alert and allows the focus of your eyes to adjust from infinity to up close continually

[APenNameAndThatA]

11 hours ago, old man emu said:

I'm concerned about a little bit of semantics here.

 

Are you using "acceleration" in the sense it means in physics - rate of change of velocity - and saying that a change in the direction or magnitude of the velocity vector is acceleration?

 

OR

Are you using "acceleration" in layman's tems of simply an increase in "speed"?

 

I have been thinking of that word in layman's terms as aircraft don't usually accelerate ( speed up or slow down) rapidly enough to greatly affect the received velocity data. That's why I say that determining acceleration would add extra steps in the computations that might not provide any better results than if it was left out.

By the time you posted this, you were perfectly well aware that RF was using the term correctly. 

Edited November 13, 2022 by APenNameAndThatA

[old man emu]

7 hours ago, APenNameAndThatA said:

By the time you posted this, you were perfectly well aware that RF was using the term correctly. 

No I wasn't. That's why I asked.

[Markdun]

In the boating world the ADSB equivalent is AIS, albeit it’s only two dimensions and the velocities are an order of magnitude less than aircraft. That said, away from congested waters like Sydney, I’ve found the AIS system fantastic. I can program my GPS/chartplotter to give me a warning if another boat is predicted to intercept my predicted position by a certain distance (say 1nm) or time. While the data from other boats does include its velocity and the first derivative (rate of turn), I’m pretty sure it just predicts the intercept based on current position and velocity( course over ground and SOG).  Further the data from other boats includes their MMSI number which enables you to call that boat’s bridge by DSC on the VHF radio. If they don’t respond you can up the urgency and initiate an alarm on the bridge. In my experience ships respond... I call them, give my position via bearing and distance and ask if they see me on their radar... they never do. Generally, I advise them I will change course to pass them port to port, and on a couple of occasions they have volunteered to speed up or slow down to give me greater clearance... this in more confined coral reefed waters. But it’s bloody useless in Sydney..... it’s just distracting noise.... there are so many potential conflicts the system just can’t cope.

 

I think the same might apply to ADSB in aircraft. In the circuit forget it.... too much else going on. But on a cross country being notified of any traffic likely to come within say 3nm or 1000’ vertically, or a specified time, would be very handy.  It’sa pity our VHF radios are so dumb and can’t use the ICAO number allocated to each ADSB transponder to call the other aircraft via a DSC channel.  And there is the problem that many aircraft monitor the incorrect frequency. Where I’m located the class G airspace ‘area frequency’ is the same as ‘Canberra approach east’. My neighbour with an autogyro only monitors 126.7 on the advice of his CFI in Wollongong and is resolute in not giving a broadcast on, or listening, to the area frequency because he thinks he’ll get into trouble with ATC. We both operate out of uncharted private airstrips.

 

BTW, organising an in air rendezvous with ADSB is amazingly easy and a little bit scary.