Wingless Walrus,
Points related to yours above:
- It is highly unlikely that a database update of crowd positions and obstacles at the display site for loading into a display aircraft could be done in enough detail and within the required timescales for it to give the risk mitigation levels that you seek.
- Minimum altitude loss requires a wings level pull-out. If an automated escape manoeuvre had to achieve a specified ground track in order to avoid the crowd or obstacles then a turn would inevitably be required to achieve this. This would increase the overall height loss and, therefore, the algorithm to decide when to initiate the manoeuvre automatically would be, I suspect, too complicated to be feasible.
- If I fly on an erect down 45 line at a safe height and fly a 360 degree roll, the automatic recovery system should not trigger. However, if I stop the roll inverted and immediately pull full back stick the high pitch rates achievable by modern fighters would almost certainly be able to beat any recovery algorithm that actually allowed the pilot to fly even safe manoeuvres and a ground impact would occur.
- When would the automatic recovery mode disengage and hand control back to the pilot? What would the power setting be at that point? If the recovery was at the g limit and the pilot was not expecting it then he/she could GLOC. Even if GLOC did not occur the pilot could be very disorientated. Disengaging is one aspect that you have not considered and I believe would be a major safety concern which could result in the system causing an accident.
- The Tornado TFR system controlled to a maximum turn rate of 7 deg/sec and left the aircraft in an open loop climb after 2 secs. This is a very limited manoeuvre envelope and not analogous to display manoeuvres.

The comments regarding a 'safety pilot' in the rear seat of the F-15 display are interesting. Without knowing the CRM aspects of how this was flown, I suspect that the rear cockpit pilot was calling out airspeeds on the way up, altitudes on the way down, and may have confirmed correct engine performance. Having displayed 2 seat aircraft, this is what my navigators did and it was very useful.

Rgds

L

Yes, it did. However, this was a rudimentary system that merely coupled a link between the operation of the (very effective) dive brakes and a pre-set elevator nose down trim bias and, a reset of those settings upon a signal from the contact altimeter. The recovery was therefore, more of a simple automated function. If the pilot mis-set the contact altimeter (baro), recovery might not be achieved...depending upon the error! :oh:

OAP

Thanks for the correction. It was a while ago. I think there was some discussion wrt GLOC but the difficulty was establishing that it was actually taking place. Some form of physiological measurement/feedback seems to be essential but you can probably imagine the aircrew response to some of the more invasive techniques :hmm:

EAP

A number of years ago we (the UK) did look at protection system that moved away from a recovery algorithm to a dynamic flight envelope protection system. The idea behind it was to prevent or frustrate any manoeuvre that could not be manually (or automatically) recovered from.

In the example you give above the system would allow the 'roll inverted' bit but limit/frustrate/prevent the worrisome pitch input. Alternatively if you had pitched-up hard first and then rolled the system would push back in roll when the lift vector became ominous. In a way it was more akin to SPILS intervention on the Tornado (physically harder to pull back with increasing AoA) but in all axes with terrain/aircraft/store limit awareness. In the sim environment you could still deliberately crash it but you had to work really hard to do so. For the Douglas Adams fans think 'throwing-yourself-at-the-ground-but-missing' - I think that line made it into the TP report....

The original idea behind the programme was to eliminate the alarms and overt interventions used with traditional GCAS whilst providing care-free handling, without ever fully-seizing control from a pilot operating at the limits of achievable flight. Without a platform customer I guess the details are archived at dstl or with whatever DEC-TA morphed into.

That systems could have saved the Italian Typhoon and pilot lost. The Typhoon already has an automatic recovery function but is pilot selected. Had your system been developed and linked to the existing auto-recovery function that would have made an effective system.

I wonder if automatic ground/air collision systems will be on the next UK fighter and will the UK develop it or buy it from the USAF?

LOMCEVAK
- the auto-GCAS appears to be a very capable ground avoidance system (as mentioned by other comments here).
It has already saved several fast jets and pilots. Some good clips on YT and good information on it in links in comment #34. Also the link below for Wright Patterson Air Force base. The USAF is also developing an ACAS (Air Collision Avoidance System).
- Automatic Collision Avoidance Technology (ACAT)
https://www.wpafb.af.mil/Welcome/Fac...chnology-acat/

The following description from The Aviationist: -

The Auto-GCAS (Automatic Ground-Collision Avoidance System) is an automatic system that, once installed and activated on a plane, monitors the flight parameters: to make it simple, when predictive algorithms and computer software determine that altitude, speed, attitude, GPS position, terrain are such that the CFIT [Controlled Flight Into Terrain] is imminent, the A-GCAS automatically (without pilot initiation) sends inputs to the flight controls to recover the plane.

Unless you have the gear down...:confused:

It would be ironic if a jet fitted with auto-GCAS crashed because it ran out of fuel because the auto-GCAS would not let the aircraft land!

I suspect they have these problems solved; they have paid particular attention to using the system in all flight regimes without unwanted behaviour. They are aiming for a very comprehensive system of collision avoidance.

Reading some of the links mentioned previously, one of the main objectives was to also avoid nuisance warnings and behaviour. It seems a very sophisticated system. It doesnt kick-in until almost the last second of decision making (I think in one event the time quoted was 0.8 sec from being too late to do anything) and the system aims for a 5G pull-out.

In essence they are building a system that replicates the pilot thought process, since the pilot is the current most sophisticated anti-collision system. At some some point the artificial intelligence will be as good as the pilot, then its time to put Cylons in the jets.

Wingless Walrus,

With your conviction for the incorporation of auto-GCAS for display flying, I am curious to know what you propose re my earlier point regarding when the system would disengage and return control authority to the pilot and the potential for it inducing GLOC.

When displaying a high performance aircraft, the aim is to demonstrate its maximum performance when safe to do so and to transition between such manoeuvres in a safe manner which, invariably, will not be flown at maximum performance. Any automatic GCAS will have to have safety margins built in, and I feel that g onset rates and peak g would have to be reduced from the ultimate capability of the aircraft such that GLOC would not be induced upon a pilot who was unprepared for the automated and unannounced high g manoeuvre. Therefore, in a display I suspect that there would be nuisance intrusions occurring when the automatic system would effectively terminate the display unnecessarily. Personally, I would not be prepared to display an aircraft that had such a system.

The key to display safety is to pick a pilot who has the correct attitude and to train him properly. This training must be done by someone with the requisite experience, not just of senior military rank. And I suspect that this paragraph may open up a whole new topic for discussion!

LOMCEVAK,
it will take someone far more knowledgeable than me to make any decision on any system to be used on any aircraft for anything. I only make the point that those far more suitable than me, i.e. the USAF and others, have gone down the path of creating an automatic ground collision avoidance system (and air collision avoidance system) that even with its current limits has saved aircraft and pilots.

I have no idea what actual parameters are used in this system for those criteria you mention. The system in its current state of development has already recovered aircraft and saved pilots that other wise would have crashed. Whatever the values of those criteria used, they are saving the aircraft and pilot. This includes cases of recovering the aircraft after the pilot has suffered GLOC.

What the system current limitations are I dont know, but it is still in ongoing development. A system goal is for use in air-air combat training/exercises to avoid ground and air collisions. That is a very dynamic and unpredictable environment. I would imagine such a system could be made of use for display flying too.
Below is a summary of your points raised previously and my thoughts.

- crowd location
Does not the pilot also need to know these in advance for the same reasons of planning the display? The Italian Typhoon was at the coast; an approximation could have been made that the beachline was the boundary between crowd and recovery area; that would have been enough to save that pilot and Typhoon.

- minimum altitude loss
Does this not also apply to a human pilot? Computers can think and act faster than a human pilot; unstable aircraft can be flown by computers that a human pilot could never control.

- rolling at 45 deg nose down
See comment #44; the current auto-GCAS appears to be able to cope with high pitch rates (system flown on F-16; see also links in comments #34 & #46); computers can sense and react faster than any human pilot, it only requires the right algorithm/method.The system makes a decision as to whether the aircraft is approaching a point from which any recovery is impossible (as a human pilot would) and then reacts at the last second to avoid it.

- When would automatic recovery mode disengage and hand control back
I dont know the criteria exactly but it is being done by the current auto-GCAS which has saved aircraft and pilots in state of with/without GLOC. The Typhoon has an auto-recovery system, pilot selected, which can recover the aircraft and hand control back to the pilot in a satisfactory manner. Automatic recovery and handing control back to pilots who have/have not suffered GLOC is successfully implemented in some current systems.

- Tornado TFR
I mentioned this only as an example of a simple ground avoidance system; the auto-GCAS is extremely sophisticated, much more than the Tornado TFR and it is still developing. Had the Italian Typhoon been fitted with the current auto-GCAS it probably would never have crashed. Current auto-GCAS may not be perfect, but it is saving lives and even a modified limited auto-GCAS could have probably saved that Typhoon, whether it was pilot error or GLOC.

I am guessing that to display pilots auto-GCAS seems the equivalent of stabiliser wheels on a bike. Auto-GCAS is being fitted to the F-35 5 years earlier than planned (from 2019). I am guessing that soon we will see an F-35 at an air display using auto-GCAS to some extent; either modified, limted or fully functional.