🚀 go-pugleaf

There are multiple manifestations of intelligent/expert control.  Here are
a few:

- control systems that use neural nets, so that the control law can be
  automatically learned rather than designed, often resulting in control
  systems that solve very hard nonlinear problems.  (Of course, then nobody
  knows whether/when the control system is stable...)
- control systems that use fuzzy rules, in part to implement human-like
  control actions, and in part as a weighted averaging of multiple control
  regions (and again not much is known about stability).
- control systems that incorporate symbolic reasoning (as in Artificial
  Intelligence) somewhere in the control loop, e.g. to decide on-the-fly
  a trajectory for a robot arm to follow through a cluttered space, or
  to plan a schedule for an automated job shop.
- control systems that incorporate symbolic reasoning which dynamically
  redesigns the control loops themselves (i.e. an extension of reconfigurable
  control).

There is an almost-annual conference on Intelligent Control, published by the
IEEE Computer Society.  The proceedings contain some flakey and even marginally
relevant stuff, but that's probably the best place to start looking.

Marcel