kite state-machine is a binary pair- Flying or Not-Flying. A workable
sensor is a contact switch lever at the anchor-point that trips when the
kiteline lays down on the surface. One nice experiment is to let a Morse
Sled land & self re-launch for extended sessions while recording state
against a time-base. The next logical elaboration of the state-machine is
to measure the wind-speed dimension while tracking kite-state. This allows
failures & latencies to be sorted from the data stream. The Beaufort Scale
is a fine default minimal granularity assumption.
From there the state machine becomes more application or operation driven.
One might care most about certain dimensions like wind direction,
altitude, kiteline state, generation, etc., but the essential task is to
minimize the number & precision of dimensions lest the state machine grow
intractable. Intractability takes many forms, but complex processing in
real-time is a common hurdle. High dimensional combinatorics burdens
processing. Sensor fusion challenges the designer. Uncertainty of every
kind limits reliable performance.
The natural wind-field is a major uncertainty. Active control is only as
good as one's wind sensing & modeling, but both engineering functions are
weak at present. A basic wind state model is around seven dimensions. More
complex versions soon become intractable.
Much work is ahead to digitize kites. A practical hint is to use
to gather experimental data in mass-parallel. Video provides an inherent
discrete time-base, enables space measurement with color-coded reference
markings, and the visual frame easily includes real-time instrument
read-outs in the foreground. Post-processing of the image stream automates
digitization. Eventually one may migrate to simpler sensing and process
fast enough to automate acceptable actuation in real-time.
A most useful single-actuation dimension (or degree-of-freedom (DOF)) to
automate is yaw output as this is useful to resist wind-veering upset
and lock-out looping. Angle of attack (AoA) is a primary control
input for power modulation and even killing, but an elastic aft-bridle can
embody the logic to reduce AoA in gusts.
Yaw, roll, and
pitch mixing are desirable for kiteplane aerobatics.
NEXT: Critical Reliability of Autonomous AWECS
FairIP/CoopIP ~~Dave Santos
May 23, 2010 Happy Birthday Otto Lilienthal M1555
Comment and development of this topic will be occurring here.
All, send notes!
Terms and aspects:
Commentary is welcome: