Blogs

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • bGatti March 26, 2010 at 4:19pm
    @Dave, sure an easystar, like most any plane, can be made stable: stable is the opposite of manueverable. If you want a plane to do 3d tricks, then the CG should be pushed back as far as possible, if you want powerful flaring at landing speeds, push the CG back, but if you want a plane that its inherently stable, then push the cg forward (and add down elevator to the trim) such a plane will, if diving, accelerate and pull up until it slows to the point where the weighted front again pulls the nose down, which leads to acceleration, and this is passive negative feedback resulting in a glide-slope that corresponds to thrust-drag. So you don't need sensors or code to establish roll or pitch stability. Yaw stability is much slower, and GPS is moderately sufficient for same.
    What a passive system will not do well is fly upwind. Passive flight cannot be stable at high speeds and low speeds, and are invariable set for a slow speed, so they can't push against much wind.
  • Dave Whittington March 26, 2010 at 3:55pm
    bGatti,
    I'm sure you are correct that it would behoove this approach to have quite a stable choice of aircraft - but perhaps an EasyStar is close enough for an eval. They have a real notion of pseudo-attitude drift compensation and it would simply be interesting to try it - if it worked well enough in certain situations (light winds,etc), just an ArduPilot board (no shields/thermopiles/IMUs, etc) with a GPS chip might be enough - seems hard to believe but why not try it!
  • bGatti March 25, 2010 at 3:50pm
    here's their thinking - it's not in code - it's in the picture.
    keep the weight under the wing - this provide automatic roll stability;
    next use a rear elevator, and weight front of wing center, which provides pitch stability.
    Viola.
  • Dave Whittington March 25, 2010 at 12:12pm
    I agree that there must be some real limitations, but wouldn't it be great to try it out if we could get their code(or at least a full set of their equations) - not for auto takeoff and landing but just for our normal autopilot fly-to-waypoints type mission. I would expect a lot of differences/problems but what is truly great about our Arduino ArduPilot environment is that "it's just code". We would make a test flight on a slightly windy day with our standard Ardupilot 2.5xx code and then their approach would be downloaded which would ignore our attitude sensors and just see what happens (BE READY TO GO TO MANUAL MODE!). Maybe we could approach them and see if they would be interested? Maybe we would learn something pretty cool that could integrate into our thinking - who knows!
  • Tj Bordelon March 25, 2010 at 7:00am
    This approach has it's limits. As Michael points out, I would wager that it only works well on calm days.
  • bGatti March 24, 2010 at 10:09pm
    @Michael
    postscript to your note. The MIT Paper was and remains that - a paper, due, I'm sure, to the failure to receive a COA, Kornfeld and company were unable to even test their theories - and if by chance, they were able to test their theories, they were prevented from taking their innovation from the ivory tower to the market because the FAA was in no mood to permit innovation - so instead of creating a new aerospace company with high paying (deficit reducing) jobs, they published a paper. A very nice paper, all white and shinny, with a very stylish font.

    The Koreans, meanwhile, more hungry for new economy opportunities, than for stifling regulations, have built a flying device that has mass and takes up space. viva la differance.
  • bGatti March 24, 2010 at 10:00pm
    The developing world has an advantage in UAV and new tech as their are fewer entrenched interests and therefore less predatory regulation. When one does simple interest-alignment analysis, it is clear (IMO) that underdeveloped countries are more interested in disruptive technologies, while in countries where industry has "taken root", there is a negative pressure on disruptive technologies:
    Take for example the coal lobby's commercial (Coal Creates Jobs) - So any energy technology (say wind, nuclear or NG) must overcome the political headwind of these existing jobs and the political chairs tied to them - even if they also "create jobs". Due to hysteresis of government (specifically congressional seniority)- old industries, and their older, more senior representatives in congress have more clout in the committees that control regulations, and thus old industries have a relative advantage over new jobs. So While Americans have a vibrant economy due to their mobile workforce; they lag behind the rest of the world in many fields: (wind, nuclear, wave energy, hybrids, batteries, trains, maglevs, healthcare per percent of GDP) Due perhaps to the perverse political interests of declining industries.
  • john bero March 24, 2010 at 9:35pm
    Integrators are designed to eliminate the steady-state errors on the velocity and position tracking errors. The aircraft speed is maintained at around 26 m/s except for takeoff and landing. Flare control uses forward speed/pitch/altitude. (twenty six meters per sec. Is that right? 62mph? This thing is a rocket ship)
  • Michael Zaffuto March 24, 2010 at 9:28pm
    Their controller is based entirely on the constraint of coordinated turns to keep turn rate and roll angle correlated. I suspect a few gusts of wind or shear from the right direction and its over.

    Not to take away from the fine implemenation from S.Korea....but....

    Let us take a moment now to bow our heads and give credit where credit is due:

    Kornfeld, Hansman and Deyst, of MIT, invented GPS Velocity Vector Based Attitude Determination, Report No. ICAT-99-5. June 1999. Contains all the theory for Single antenna GPS based pseudo-attitude and psuedo-flight angle(pitch) determination.
  • Developer
    Mark Colwell March 24, 2010 at 8:15pm
    Based on good old American software technology,

    credits from article:
    Manufacturers

    The reference station and UAV on-board receiver were both the 3151R from Novatel, Inc., Calgary, Alberta, Canada. The software for both onboard and ground stations was developed using Visual C++ with MFC, which runs on the Windows 2000/XP Professional operating system from Microsoft Inc., Redmond, Washington, USA. The data was plotted by using MATLAB from Mathworks, Inc., Natick, Massachusetts, USA.
    Copyright © 2010 Gibbons Media & Research LLC, all rights reserved.
This reply was deleted.