Main Menu

Boeing new product reveal

Started by Spam, December 15, 2017, 07:27:11 PM

0 Members and 1 Guest are viewing this topic.


Live2Learn

Interesting concepts.  A foreseeable large weakness would be the probable imperative for the 'swarm' to protect the 'queen'... the crewed aircraft.  It makes more sense to me for the swarm to be fully AI managed with a remotely located command center.  I expect that might be a tough technology break from past human centered weapons systems to other centered.  Dunno if this model is a direct response to the shortage of flesh & blood pilots, probable inability of human centered weapons systems to successfully compete with machine centered systems, or both. 

PHall

Not having to support and protect a human pilot results in a large weight savings.
Lighter weight = Longer Range / Increased Payload. Both good things.

Live2Learn

#3
Quote from: PHall on December 16, 2017, 03:39:37 PM
Not having to support and protect a human pilot results in a large weight savings.
Lighter weight = Longer Range / Increased Payload. Both good things.

All of which = significant performance benefits.  Plus, it's easy to imagine, and therefore anticipate, an AI flown aircraft in the near future pulling 25 'G in combat maneuvers -- which would turn a human pilot into red goo...  :O

CAPLTC

Quote from: Live2Learn on December 16, 2017, 02:58:21 PM
Interesting concepts.  A foreseeable large weakness would be the probable imperative for the 'swarm' to protect the 'queen'... the crewed aircraft.  It makes more sense to me for the swarm to be fully AI managed with a remotely located command center.  I expect that might be a tough technology break from past human centered weapons systems to other centered.  Dunno if this model is a direct response to the shortage of flesh & blood pilots, probable inability of human centered weapons systems to successfully compete with machine centered systems, or both.

It is in response to the reality of AI...
For now a human must remain in the loop for moral and legal reasons but we'll soon get past that and onto fully-autonomous killing.
"Find the enemy that wants to end this experiment (in American democracy) and kill every one of them until they're so sick of the killing that they leave us and our freedoms intact." -- SECDEF Mattis

PHall

Quote from: CAPLTC on December 17, 2017, 03:29:23 PM
Quote from: Live2Learn on December 16, 2017, 02:58:21 PM
Interesting concepts.  A foreseeable large weakness would be the probable imperative for the 'swarm' to protect the 'queen'... the crewed aircraft.  It makes more sense to me for the swarm to be fully AI managed with a remotely located command center.  I expect that might be a tough technology break from past human centered weapons systems to other centered.  Dunno if this model is a direct response to the shortage of flesh & blood pilots, probable inability of human centered weapons systems to successfully compete with machine centered systems, or both.

It is in response to the reality of AI...
For now a human must remain in the loop for moral and legal reasons but we'll soon get past that and onto fully-autonomous killing.

AI has a ways to go before it's ready to go solo. The current technology doesn't handle unexpected events and conditions too well, yet...
For now the Mark I, Mod 0 Human Brain is still superior.

Spam


I would say that there are indeed some mission sets and conditions where automation "handle unexpected events and conditions" far better than humans. Examples are algorithmic sensor-based responses to envelope, signature, or terrain envelope protection (e.g. auto GCAS).  For operators who are faced with multiple UAS operations (perhaps while also flying a manned platform, within a networked environment, with net enabled weapons) the challenge is mating a proper mission/task analysis to a solid automation design to have a failure tolerant/redundant system, within tactically significant engagement timelines.

Here's an abstract from a paper I gave this fall at a DC area conference (the design approach we are taking is based on the need for an automation Executive to arbitrate between individual automation policy managers, such as AGCAS, signature managers, weapon envelope managers, fuel policy managers, etc.):


"(U) Abstract
Despite decades of industry experience in the design of Unmanned Aerial Vehicle (UAV) control systems and their user interfaces, a combination of factors persist that produce a significant and unacceptable loss rate of UAVs due to poor user interfaces. One significant element is the current focus of human systems design on lower-order User Interfaces (UI) at the expense of investing in the design of an adaptive higher level integration to relieve inattentive or overtaxed operators of significant functionality as required, and to perform time-critical tactical tasks which humans cannot perform or for which they are not well suited. The approach proposed is one which defines the respective roles of user interactions with adaptive policy manager automation to address the loss of vehicles and mission failures. Specific policy manager automation elements are explored which will enable the system to flexibly assume or release UAV vehicle or systems functionality based on operator action/saturation in a number of mission areas. A notional Executive automation controller design approach is outlined to meet time critical information integration and mission task requirements".

V/r
Spam



Live2Learn

Quote from: Spam on December 17, 2017, 09:17:57 PM

"(U) Abstract
Despite decades of industry experience in the design of Unmanned Aerial Vehicle (UAV) control systems and their user interfaces, a combination of factors persist that produce a significant and unacceptable loss rate of UAVs due to poor user interfaces. One significant element is the current focus of human systems design on lower-order User Interfaces (UI) at the expense of investing in the design of an adaptive higher level integration to relieve inattentive or overtaxed operators of significant functionality as required, and to perform time-critical tactical tasks which humans cannot perform or for which they are not well suited. The approach proposed is one which defines the respective roles of user interactions with adaptive policy manager automation to address the loss of vehicles and mission failures. Specific policy manager automation elements are explored which will enable the system to flexibly assume or release UAV vehicle or systems functionality based on operator action/saturation in a number of mission areas. A notional Executive automation controller design approach is outlined to meet time critical information integration and mission task requirements".

V/r
Spam

Interesting abstract.  It describes an important interim step in the evolution of AI.  How far out is the edge of the current research envelope for these systems?

Spam

Depends on the customer, the mission, the system, and the user population. USAF, the USN, and the Army have distinctly differing approaches to user selection and training, even when equipment is similar (MQ-1/MQ-9) and mission sets overlap. Commercial foci of research diverge even more, and is more risk-tolerant in pushing design concepts. Multi agent adaptive automation isn't new (JSF was designing policy managers in the 90s) but with multiple UAVs doing different and sometimes cooperative things, the design of an executive was the focus of our work.

Ready for the reveal, now?

V/r
Spam