I was thinking along those lines, yes. jan.
On Tue, Aug 23, 2011 at 10:34 PM, Dave <e...@dc9.tzo.com> wrote: > I tried a first and second order lag filter and I didn't like the > results.. I threw a simple averaging filter at it and it looks > "intuitively" good and it is super simple. > > Hopefully good enough. > > We will see if the boat agrees. :-) > > The pursuit of some fuzzy logic might make sense if the averaging filter > is inadequate by itself. > Perhaps some fuzzy logic altering the averaging filter on the fly. > > Dave > > On 8/23/2011 3:47 PM, Jan de Kruyf wrote: > > Why all so terribly deterministic. > > Throw some fuzzi logic at it. > > Very good for changing a slowly changing error. > > > > jan. > > > > On Mon, Aug 22, 2011 at 10:16 PM, Dave<e...@dc9.tzo.com> wrote: > > > > > >> I know... but they have a C compiler for those boards... I don't have > >> that luxury.. maybe in the next iteration if this works out. > >> > >> Thanks, > >> > >> Dave > >> > >> On 8/22/2011 12:16 PM, Michael Haberler wrote: > >> > >>> the folks building quadrocopters use these filters a lot - on quite > >>> > >> limited CPU's > >> > >>> -m > >>> > >>> Am 22.08.2011 um 15:24 schrieb Dave: > >>> > >>> > >>> > >>>> Keith, > >>>> > >>>> The Kalman filter is probably too difficult to calculate on the > >>>> controller I am using (a micro PLC with floating point math) , so I am > >>>> going to do a first and perhaps a second order lag filter first and > see > >>>> what happens. > >>>> If the lag filters are in-effective then I will chase a Kalman filter > >>>> solution. I am going to try out your example anyway since it sounds > >>>> really interesting. :) > >>>> > >>>> Thanks, > >>>> > >>>> Dave > >>>> > >>>> On 8/21/2011 3:25 PM, Keith Curtis wrote: > >>>> > >>>> > >>>>> Hi Dave, > >>>>> > >>>>> I have a tiny bit of experience with Kalman filtering for state > >>>>> estimation, and attached a toy example that might help get you > >>>>> started. I have written it to work with Octave which is open source > so > >>>>> you can install it if you don't already have it. The idea as I > >>>>> understand it is that you have a model of how the state updates, like > >>>>> Newton's laws. You predict what you'll measure, compare that to your > >>>>> actual measurement and the apply the Kalman gain matrix K to the > >>>>> difference to update your estimate of the state. > >>>>> > >>>>> You can run the example in Octave with > >>>>> > >>>>> octave:1> kalman_example > >>>>> > >>>>> In the example, I have an object moving in a parabolic arc, and in a > >>>>> commeted-out line, an object moving in a straight line. The > >>>>> measurements that are the input the filter are stored in y. You can > >>>>> plot the measurements out with > >>>>> > >>>>> plot(y(1,:), y(2,:), 'o') > >>>>> > >>>>> At the end of the script the estimate state positions are plotted > with > >>>>> > >>>>> plot(x_hat_plus(1,:), x_hat_plus(2,:), 'o') > >>>>> > >>>>> > >>>>> The tricky parts seem to be what sort values to put in the initial P, > >>>>> Q, and R covariance matrices. > >>>>> > >>>>> Hope this helps. > >>>>> > >>>>> > >>>>> Cheers, > >>>>> Keith > >>>>> > >>>>> On Sun, Aug 21, 2011 at 12:08 PM,<kth.cur...@yahoo.com> wrote: > >>>>> > >>>>> > >>>>> > >>>>>> ---------- Forwarded message ---------- > >>>>>> Date: Wed, 17 Aug 2011 16:18:51 -0400 > >>>>>> From: Dave<e...@dc9.tzo.com> > >>>>>> Reply-To: "Enhanced Machine Controller (EMC)"< > >>>>>> > >> emc-users@lists.sourceforge.net> > >> > >>>>>> To: emc-users@lists.sourceforge.net > >>>>>> Subject: [Emc-users] OT ? Kalman Filter > >>>>>> > >>>>>> > >>>>>> I have an application that I believe could greatly benefit from a > >>>>>> > >> Kalman > >> > >>>>>> Filter. Kalman filters are oftentimes used for guidance systems > and > >>>>>> that is what I need it for - except that I am guiding a boat! > >>>>>> However the same filter I understand can also be used to predict > >>>>>> (estimate) velocity and position based upon what would normally be > >>>>>> considered incomplete information. One example I heard what that a > >>>>>> Kalman filter can > >>>>>> be used to predict velocity when used with low count encoders - as > in > >>>>>> low count encoders used on servo motors. > >>>>>> > >>>>>> If you do a look up of "Kalman filter" on Youtube you can get an > idea > >>>>>> > >> > > > > ------------------------------------------------------------------------------ > Get a FREE DOWNLOAD! and learn more about uberSVN rich system, > user administration capabilities and model configuration. Take > the hassle out of deploying and managing Subversion and the > tools developers use with it. http://p.sf.net/sfu/wandisco-d2d-2 > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > ------------------------------------------------------------------------------ EMC VNX: the world's simplest storage, starting under $10K The only unified storage solution that offers unified management Up to 160% more powerful than alternatives and 25% more efficient. Guaranteed. http://p.sf.net/sfu/emc-vnx-dev2dev _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
