RAF Mosquito

[ORAC]

https://www.thetimes.co.uk/article/3...ones-k0cdk8lds

£30m deal to build Britain’s first fleet of unmanned fighter drones

Britain’s first fleet of unmanned fighter aircraft will be developed in Northern Ireland following a £30 million investment by the Ministry of Defence.

The “loyal wingman” aircraft, as it has been nicknamed, will be designed to fly at high speeds alongside fighter jets such as the Typhoon or F-35.

The contract to design and manufacture the prototype, which is expected to support 100 jobs, was handed to Spirit AeroSystems, an American company, in Belfast in a three-year deal.

Team Mosquito will develop the RAF’s Lightweight Affordable Novel Combat Aircraft (Lanca) technology, with a vehicle flight-test programme expected by the end of 2023. The MoD plans to start manufacturing the first aircraft by 2025.

Armed with missiles, surveillance and electronic warfare technology, it will be Britain’s first uncrewed aircraft able to target and shoot down enemy aircraft and survive against surface-to-air missiles. The crewed aircraft will be able to assign tasks such as electronic warfare, surveillance or bombing missions to the Lanca drones, increasing the air power at a lower cost and risk to RAF aircrew.......

Richard Berthon, director of the MoD’s Future Combat Air, said the RAF’s Lanca technology was a vital element of its future combat capabilities. “Autonomous ‘loyal wingman’ aircraft create the opportunity to expand, diversify and rapidly upgrade combat air forces in a cost-effective way, now and in the future,” he said.

Mike Wigston, chief of the air staff, said: “We’re taking a revolutionary approach, looking at a game-changing mix of swarming drones and uncrewed fighter aircraft . . . alongside piloted fighters like Tempest, that will transform the combat battlespace in a way not seen since the advent of the jet age.”......

[Wensleydale]

So as well as combat it can perform ISTAR roles as well? Why not call it the Lightweight Affordable Novel Combat Identification Surveillance Target Acquisition and Reconnaissance, or "LANCISTAR" instead of Mosquito?

I'd like to see a fast jet pilot - in combat, flying both his own aircraft and a swarm of drones.......

[Bob Viking]

Wensleydale

The whole point of AI is that it is pre-programmed to know what to do and follow a bunch of pre-determined parameters.

If it is well designed and well integrated the FJ pilot won’t need to do very much at all. The drones are just there as assets at his/her disposal. Like having extra missiles or a longer range sensor.

I really don’t think it will involve a pilot flying a Typhoon/F35/Tempest whilst simultaneously controlling a bunch of drones with a controller!

BV

[orca]

Spot on BV. And quite aside from AI an autonomous system can be wonderfully dumb - in a positive sense. No AI whatsoever required for the following:

Follow the ACO outbound to the millimetre.
Don’t run out of gas.
Don’t transgress any known ACM.
Don’t penetrate a known MEZ. (I’ll let you know by text if there are any new ones).
Don’t leave the FIR.
Don’t fly into me.
Don’t get more than 20nm from me.
Don’t employ without my say so.
If you’re near a POI and can capture an ISR product please do.
If we lose comms for X minutes go back to CAP, for Y minutes then RTB.
Go home obeying the ACO to the millimetre.

All things you might hope your manned wingy would achieve but that you could never guarantee - now simply check boxes on a mission planner.

[ORAC]

All things you might hope your manned wingy would achieve but that you could never guarantee - now simply check boxes on a mission planner.

You Tube

You Tube

[just another jocky]

Originally Posted by Bob Viking

The whole point of AI is that it is pre-programmed to know what to do and follow a bunch of pre-determined parameters.

If it is well designed and well integrated the FJ pilot won’t need to do very much at all. The drones are just there as assets at his/her disposal. Like having extra missiles or a longer range sensor.

I really don’t think it will involve a pilot flying a Typhoon/F35/Tempest whilst simultaneously controlling a bunch of drones with a controller!

BV

Hopefully that will stem all the usual naysayers who think we just can't manage without their insight since they left.

[ORAC]

Mike Wigston, chief of the air staff, said: “We’re taking a revolutionary approach, looking at a game-changing mix of swarming drones and uncrewed fighter aircraft . . . alongside piloted fighters like Tempest, that will transform the combat battlespace in a way not seen since the advent of the jet age.”......

You Tube

You Tube

[safetypee]

Voice activated, secure data linked - 'echelon starboard go'; 'diamond nine go'; … …
Advanced formation flying.

You Tube

Basic training

You Tube

[Wensleydale]

At last - the movement of the Red Arrows to Waddington makes sense. One jet, eight drones and Red 1 flies his aerobatics routine with everything else on pre-programmed automatic pilot.

[Timelord]

Originally Posted by Bob Viking

The whole point of AI is that it is pre-programmed to know what to do and follow a bunch of pre-determined parameters.

If it is well designed and well integrated the FJ pilot won’t need to do very much at all. The drones are just there as assets at his/her disposal. Like having extra missiles or a longer range sensor.

I really don’t think it will involve a pilot flying a Typhoon/F35/Tempest whilst simultaneously controlling a bunch of drones with a controller!

BV

s
Indeed. According to Project Astra it will be so easy to fly manned and unmanned vehicles that “Other professions” will do it in their spare time.

[just another jocky]

Originally Posted by Timelord

s
Indeed. According to Project Astra it will be so easy to fly manned and unmanned vehicles that “Other professions” will do it in their spare time.

If technology could make it easy to fly, can you think of a reason why they shouldn't? Or should we purposely make aircraft difficult to fly?

Give me easy every day, then I can focus on the job at hand.

[Timelord]

No reason at all. My doubt is about operating them being a part time duty for “Other professions”. Astra seems to envisage your “Job at hand” rather differently.

[ORAC]

https://www.thedrive.com/the-war-zon...es-jet-emerges

New Unmanned Loyal Wingman Design Based On Stealthy “Son Of Ares" Jet Emerges

Scaled Composites, together with its parent company Northrop Grumman, has unveiled a new unmanned aircraft design, known as the Model 437, which could be configured as a "loyal wingman" intended to work networked together with manned platforms. As such, it could meet the requirements of various efforts seeking to develop, at least in part, this kind of capability, such as the U.S. Air Force's Skyborg program and the U.K. Royal Air Force's Project Mosquito. The Model 437 is also notably derived from earlier stealthy Model 401, also known as the "Son of Ares," which Scaled Composites has now confirmed is intended to be pilot-optional.......

Twitter

[t43562]

You Tube

This is a long video but worth listening to IMO. I've put it at a point where you might get a sense of the state of the game in AI. Prof Furber is one of the creators of the ARM instruction set that's powering all your phones and he is now working on a project that has built a neuromorphic computer at Manchester University which has 1 million processors with an architecture that's designed for emulating spiking neural networks.

Elsewhere in the video you'll hear how a room--sized machine could now emulate a mouse brain if we knew enough about how it works to create a model.

This is not to be down on AI at all but I feel that those of us outside the field might not have a very good idea of the difficulty of the challenge.

[ORAC]

But they’re not looking at AI, just enough algorithmic ability to target independently and perform air combat.

https://www.darpa.mil/news-events/2020-08-26

https://www.defensenews.com/air/2021...ongshot-drone/

[Arclite01]

I am assuming they are cheaper to purchase.................

We are going to need large numbers of these and places to store them. I assume they launch from a runway too.

And large numbers of trained technicians to maintain what is in essence an airframe.

It'll be interesting to see these develop.

Arc

[ORAC]

We are going to need large numbers of these and places to store them. I assume they launch from a runway too.

Those such as Longshot are planned to be able to be air launched by the B-21 and perhaps underwing my other platforms. They could also possibly by ground launched by truck based launchers - g forces not being an issue.

The advantage of such a platform is that it doesn’t need to practice to remain current so no training flights are required and they can be stored in hermetically sealed containers for prolonged periods negating the need for maintenance with software updates and power provided by umbilicals if necessary.

Build a version with folding wings and control surface and they could even be container launched by ships and submarines - a major factor in a region with long transit times such as the pacific.

[t43562]

Originally Posted by ORAC

But they’re not looking at AI, just enough algorithmic ability to target independently and perform air combat.

https://www.darpa.mil/news-events/2020-08-26

https://www.defensenews.com/air/2021...ongshot-drone/

I did read about that and I'm sure that impressive things will get done but I just caution that demos like this often include "simplifications" of one kind or another that don't get emphasised in the tests and billions have been and will be spent on things that have horrible behavior in edge cases. While you're busy thinking how amazing it is you should remember that it might be about as smart as e.g. a bee - if we're lucky. That leaves a *lot* of possible situations which it might not be able to handle. This is not forever - because the technology is advancing - it's just that it gets hyped so much that one tends to lose all sense of proportion. There is a debate you could have about Algorithmic approaches versus machine learning or deep neural networks but as I understand it many of the cool things that can be done (such as identifying things based on signals or images) will require ML or Neural nets and for these it is not easy to prove that there is no odd or unexpected behavior caused by biases in the training data. So I think it's still more the case that we'll be using humans-with-ai-assistance for a while. I'm not important in the field or anything like that all so FWIW my opinion doesn't carry any weight.

[Arclite01]

Originally Posted by ORAC

Those such as Longshot are planned to be able to be air launched by the B-21 and perhaps underwing my other platforms. They could also possibly by ground launched by truck based launchers - g forces not being an issue.

The advantage of such a platform is that it doesn’t need to practice to remain current so no training flights are required and they can be stored in hermetically sealed containers for prolonged periods negating the need for maintenance with software updates and power provided by umbilicals if necessary.

Build a version with folding wings and control surface and they could even be container launched by ships and submarines - a major factor in a region with long transit times such as the pacific.

That is actually really interesting - especially the storage aspects...... not sure we will ever have a platform to air launch ours though.

Arc

[Lookleft]

The “loyal wingman” aircraft, as it has been nicknamed, will be designed to fly at high speeds alongside fighter jets such as the Typhoon or F-35.

I know you Englishman think that anything worthwhile can only be built and operated by HM subjects but get onboard or get out of the way...if you wouldn't mind awfully.

https://www.airforce.gov.au/our-mission/loyal-wingman