Unmanned dogfight logic


I never believed the widespread assertion that air combat is too difficult and complex, so manned fighters are unavoidable.* (I'm nevertheless not convinced that unmanned combat aviation is the way to go, except for extreme aircraft; small size, high endurance or high altitude).

Now there's some substantial support for this view:



Especially defensive (dodging) manoeuvres against incoming missiles are almost guaranteed to become automated (though triggered by pilot) once the necessary 360°x360° sensors have become common. That will likely happen in the 2020's, as systems such as the DAS can be retrofitted to existing aircraft (the typical missile warning sensors of today aren't as capable).

Slightly related:



*: Air forces are largely run by generals who used to be (fighter) pilots, so the 'experts' in the armed services have a  conflict of interest and the retired experts are emotionally compromised and thus biased on this issue. A pilotless air force would be more than a technological shift; it would be a cultural and organisational break. Air forces would also lose much budget, much personnel, much prestige and many slots for officers if fighter wings were turned into missile battery-like organisations that do not need to have flight training on most days of the year.


  1. Most AI claims should be taken with a healthy dose of skepticism. Kasparov being beaten by AI was supposed to mean that AI are superior to human players. Yet 20 years latter, they are not.

    Usually, there are two ways such result happen,
    a) cheating (with human interference to offset the AI limitations)
    b) the opponent gets paid to lose.

    The AI could be hard pressed to not always use the weapons in a certain pattern for example. Once the human understands that logic, he will be able to exploit it.

    Dodge manouvers on command are not really AI.

  2. What would the advantage of automated dodging maneuvers be?

    1. Less errors, more precision and full exploitation of the aircraft's manoeuvring potential regardless of whether the pilot stays conscious (and seeing) or not. Furthermore, the technology could keep staring at the incoming missile while the pilot could hardly look at much else than forward, up, left or right - and not particularly well so at 7+ g.
      The automated system could also see the missile at night and through some clouds, while the pilot could do so only with help of a helmet-mounted display.