One for the future perhaps?

[Jimbuna]

UK Taranis unmanned combat aircraft demostrator likely basis for post-2030 UK airforce.

http://nextbigfuture.com/2015/01/uk-...-aircraft.html

[Onkel Neal]

Nice,

[Oberon]

She's a sexy looking beast the Taranis, can't deny it.

[Wolferz]

Boomerangs from Australia would likely cost less.

[STEED]

likely basis for post-2030 UK airforce

Yea right.







Jim been UFO spotting again.

[Oberon]

Well, it's fair to assume that a good number of the 'triangular' shaped UFOs spotted in the 1980s/1990s was a certain bomber.

[Jimbuna]

True that

[ikalugin]

I think that strike UAVs are over rated, as far as you won't have a proper AI, you won't be able to hand over targeting and weapons use to it.

[Oberon]

Originally Posted by ikalugin

I think that strike UAVs are over rated, as far as you won't have a proper AI, you won't be able to hand over targeting and weapons use to it.

I think it would be a mistake to give a strike UAV total AI control, at this current level of AI intelligence anyway, no telling what it might bomb. The best bet is for the UAV to operate on AI control until it notices and designates a target of interest, then it notifies its human controller who can make the decision on whether to prosecute the target or not. That way you can have one man controlling a squadron of UAVs who will be able to be tasked with attacking different targets with minimal human oversight. In a way, the human should be the safety, the go/no-go element of the UAV. It's not fool-proof, humans make mistakes too, probably more so than machines, but I think putting total control into AI hands is...quite honestly perhaps a little too risky right now, and possibly too risky in the future too. Skynet and all that.

[CCIP]

Originally Posted by Oberon

I think it would be a mistake to give a strike UAV total AI control, at this current level of AI intelligence anyway, no telling what it might bomb. The best bet is for the UAV to operate on AI control until it notices and designates a target of interest, then it notifies its human controller who can make the decision on whether to prosecute the target or not. That way you can have one man controlling a squadron of UAVs who will be able to be tasked with attacking different targets with minimal human oversight. In a way, the human should be the safety, the go/no-go element of the UAV. It's not fool-proof, humans make mistakes too, probably more so than machines, but I think putting total control into AI hands is...quite honestly perhaps a little too risky right now, and possibly too risky in the future too. Skynet and all that.

The problem that I foresee getting in the way of this, though, is that this means reliable communications. Unfortunately, at least against sophisticated enemies, that is potentially very prone to being jammed.

[Armistead]

You send 20 or so out, the enemy jacks your software, takes over, turns them back on you.....

[ikalugin]

The issue, as it was already mentioned is the communications for any remotely controlled UAVs.

The reason stems from simple physics and the way the jam resistance works now a days. For a remotely controlled vehicle you need to send relatively large volumes of information.

That means that you use up your bandwidth, which precludes effective frequency hopping and other such techniques.

[CCIP]

Originally Posted by Armistead

You send 20 or so out, the enemy jacks your software, takes over, turns them back on you.....

I honestly don't think that's a huge threat in itself - I'm fairly positive that cyber-security is going to improve by a lot in that regard. On the other hand, I don't think it's ever going to be possible to have wireless communications that are not going to be susceptible to jamming and/or EMP. That doesn't require a lot of finesse - just an emitter that's powerful enough.

[Oberon]

Originally Posted by ikalugin

The issue, as it was already mentioned is the communications for any remotely controlled UAVs.

The reason stems from simple physics and the way the jam resistance works now a days. For a remotely controlled vehicle you need to send relatively large volumes of information.

That means that you use up your bandwidth, which precludes effective frequency hopping and other such techniques.

The key really would be to minimize the comms traffic between drone and operator by using AI for basic operations, take-off, flying, target search and then burst traffic communication to the operator basically saying "I found this, can I kill it?" The drone then follows the target until the operator says "Yes" in which case it deploys a Hellfire, or "No" in which case it abandons that target and searches for another. In regards to jamming, in that case the best bet would be to have the drone abort and RTB if a ping back from HQ isn't received in a set time.

[ikalugin]

That would still require sending a video feed (or a video file) over in a timely fashion, which again assumes a broadband chanel.

Meanwhile: