I wonder if the operator who will use MIUS will be able to use this UAV inside a 360-degree training simulation which has environmental awareness (HAVELSAN tech.) within and engage in a dogfight or carry out an operation? It would be marvelous but is it possible technological wise.
Only visuals or representative symbology without motion (it's better not to stress human body) should be enough for an expert human to dogfight or to perform any kind of operation. But if we have too many UCAVs there would be bandwidth limitations. So there would be a limit to how many UCAVs can be controlled like this.
Instead of this we could offload low level functions to AI, like flying or dogfighting and simple decision making (i.e. use terrain to get away from SAM threat) and control high level functions like decision making i.e which areas should be attacked, rules of engagement, weapon launch confirmations to human operators. In time we can offload this kind of high level decision making to AI too.
Alternatively manned fighters and other aircraft can control those UCAVs too. An optionally manned UCAV or unmanned fighter aircraft would do as well.
The most probable threat we'll face is datalink jamming and hijacking of unmanned vehicles. The second would be attacks against AI logic. AI is still stupid like other computer software, on top of that most AI methods perform like a black box and people do not understand what they do. If the enemy discovers a method to fool it he will exploit it. A country like Turkey which does not have deep roots in information technologies and cybersecurity would be extremely vulnerable.
Sample attacks agains AI image recognition:
New research could make deep-learning models much harder to manipulate in harmful ways.
www.technologyreview.com
In this tutorial, you will learn how to break deep learning models using image-based adversarial attacks. We will implement our adversarial attacks using the Keras and TensorFlow deep learning libraries.
pyimagesearch.com
The other threat would be reduction of human expertise. If AI does the job good enough we would lack pilots who can do the same job in a better way.
So if I was the state my roadmap would be:
- mass production of unmanned material using state of the art production techniques.
- produce i.e 30-40 unmanned vehicle in a month instead of below 10
- produce them from many factories including factories at battlefield
- make components cheaper and cheaper
- program simple swarm logic to those platforms (like ant colonies individual drones would be very basic but what they do as a hive would be very complex)
- offload high level decision making to specialized units or humans
- do not let humans to lose their edge - i.e. put pilots to VR and real dogfights and to operations 24/7
- make it national duty to control & design those kind of machines i.e. video games to practice or to find new methods
In the short term we will be getting limited by engines to power these air vehicles for mass production.
Red Apple?]
