William Premerlani
  • Male

William Premerlani's Friends

  • Michael Sierra
  • Mikhail M. Paremski
  • Patrick O'Neill
  • xuzeyang
  • Pawel Korzec
  • Rolf Weinzierl
  • weiqiang
  • Leonardo Garberoglio
  • Charles Radley
  • baixiaoxiao
  • Павел Ардашов
  • Zheming Wu
  • Mr.Cruis
  • bahloul

William Premerlani's Groups

William Premerlani's Discussions

Centrifugal compensation with only gyros and accelerometers

Started this discussion. Last reply by William Premerlani Jul 6, 2016. 12 Replies

AMA rule 9

Started this discussion. Last reply by Duane Brocious Nov 5, 2010. 15 Replies


William Premerlani's Page

Latest Activity

Pete Hollands commented on William Premerlani's blog post Acceleration compensation: flight testing
"Acceleration compensation in the Earth Frame  was removed from MatrixPilot trunk (Master on Github) on 18th May 2013."
Mar 22

William Premerlani replied to William Premerlani's discussion Centrifugal compensation with only gyros and accelerometers
"Mark, Thank you so much for your time and effort in testing this concept in multicopters. It performed better than I expected. It is my conjecture that the combination of your expert piloting skills and the performance of the feedback controls (I…"
Jul 6, 2016
Mark Whitehorn replied to William Premerlani's discussion Centrifugal compensation with only gyros and accelerometers
"Hi Bill,  I finally got around to playing with this concept in multicopters, and it's looking very encouraging for estimating airspeed. Here's a plot from a quadcopter flight log where I flew circles in a "coordinated" way,…"
Jul 5, 2016
johnson replied to William Premerlani's discussion robust estimator of the direction cosine matrix
"Bill  thanks very much . I got it. that's very helpful. thanks"
May 1, 2016

William Premerlani replied to William Premerlani's discussion robust estimator of the direction cosine matrix
"Johnson, There are several methods for using a magnetometer for gyro drift compensation. Here is the method that I like to use: 1. Transform the 3D measured magnetic field vector from body frame to earth frame. 2. Take the cross product of the…"
Apr 26, 2016
johnson replied to William Premerlani's discussion robust estimator of the direction cosine matrix
"Dear Bill "Magnetometer is used for yaw compensation." I think there is little mentioned about this approach in your DcmDraft paper would you please give more information about how to compensate the yaw correction by Magnetometer? Thanks"
Apr 25, 2016
johnson replied to William Premerlani's discussion robust estimator of the direction cosine matrix
"Very sorry to trouble you so many times. I have understood and let me answer some of the question myself. >question 2: >we take the cross product Z row of estimated DCM with the gravity reference vector >and if the result is not ZERO, you…"
Apr 22, 2016
johnson replied to William Premerlani's discussion robust estimator of the direction cosine matrix
"Additional question regarding the"
Apr 22, 2016

Profile Information

About Me:
I am the designer of a UAV controller with GPS, 3 gyros, and 3 axis accelerometer, being sold by Spark Fun Electronics.
Schenectady, New York

William Premerlani's Videos

  • Add Videos
  • View All

William Premerlani's Blog


Helical turn controls

Posted on June 18, 2015 at 5:00pm 4 Comments

First of all, thanks to Peter Hollands, who produced the above picture of a helical flight trajectory from one of his HILSIM test flights of the latest version of fixed wing flight controls for the open source MatrixPilot project, also known as the…



MatrixPilot precision navigation algorithms

Posted on June 20, 2013 at 1:49pm 17 Comments

The above flight track highlights the performance of the latest version of the navigation algorithms in MatrixPilot. Flight tracks are from a flight that Peter Hollands conducted. (Thank you, Peter.) The red and the green tracks represent flight segments in opposite directions between the same pair of…



global waypoint range for MatrixPilot

Posted on November 23, 2012 at 7:37pm 1 Comment

UAVDevBoard users:


MatrixPilot trunk, r1817 now has global waypoint range, provided that:

1. You use absolute waypoints. Specify longitude, latitude, and altitude for each…


Acceleration compensation: flight testing

Posted on May 1, 2012 at 9:30am 21 Comments


I have recently performed flight tests on a major change in the way acceleration is accounted for in performing roll-pitch drift compensation from accelerometer measurements. The theory behind the idea was recently published in a…



Measuring Tilt Dynamics of Multicopters

Posted on March 4, 2012 at 6:00pm 29 Comments


Presently I am doing work on multicopter modeling and control. One topic of interest to me is how can the tilt dynamic model of a multicopter be accurately determined from flight data, without disabling the controls or injecting any sort of intrusive signals. It turns out it can be done from normal flight…


Comment Wall (80 comments)

At 7:57am on February 28, 2008,
3D Robotics
Chris Anderson

Welcome! Your board and especially your documentation are VERY impressive. Would you like to do another one of our Q&A interviews here to explain to everyone your thinking behind developing the board and some of the things people have done with it?


At 4:24pm on January 7, 2009, Nathan said…
Bill, I'm interested in using your older 2-axis board for an application. Do you know of any discussions dedicated to it? Awesome documentation, BTW. Having thought myself for long hours on how to do all of that in PIC assembly, I know what a monumental task it must have been.Thanks.
At 11:06am on April 27, 2009, Tom Yochum said…
Hi Bill,
I am very impressed with the work you have accomplished. My own project is in many ways an effort to duplicate your results (cheap IMU as an AHRS for micro-UAV control). I am currently tackling the problem of initial alignment of the AHRS. I have skimmed through the DCM paper and did not see anything there. I haven't looked through your code yet either. Could you give me a quick idea of how you did the initial alignment?

At 6:13pm on July 29, 2009, Roy Brewer said…
Thanks for getting back to me with this. This would be great news if it pans out. Let me know.

BTW, would that be 6 HW PWM outputs, or would some of those be in s/w?

- Roy
At 6:52pm on September 21, 2009, Justin said…
Hi Bill,

My apologies...this has been my fault all along. I assumed I had the wires connected to the board correctly becaues I was measuring a positive voltage between the outside pin and the middle pin. I didn't think I would be able to detect a voltage from the signal.

In your previous comment when you said that I should measure a voltage with respect to ground for the signal I went back and did some futher testing and realized I simply had them plugged in backward.

The RollPitchYaw demo is working great now! I'll load up John's heli code and give that a test as well with the receiver properly connected and I bet it will work just fine.

Thanks again!

At 7:41pm on September 28, 2009, Frank Hermes said…
Hi Bill,

No I do not mind (publishing), but I appreciate your asking - as you can tell, I am very ignorant in this area, and humbled when i read through all the guys out there that are so knowledgeable, but doing my best to learn...I appreciate your patience...

You know, I thought after reading your paper on DCM that I was right back to where I was in regards to the gravity thing and the accelerometers, but since you addressed that aspect regarding using them anyway to correct the gyro drift that maybe I once again just didn't understand - I thought somehow since you were collecting continuous data that you were able to integrate the accelerometer only (no appreciable gravity influence) data into the gyro data - yikes, WRONG, again!

I would appreciate any thoughts from you or readers...this thing is really starting to bug me for a solution (reasonable one!) and I am willing to invest as long as I know there is some chance of success, particularly in regards to the COS since all I need there is data output - obviously the "canarderon" thing will take some mechanical magic as well...

I really do appreciate your time and thoughts, Frank
At 11:39am on September 30, 2009, Frank Hermes said…

Hi Bill...
I have a graph taken from one of the simulation programs we use to design the rockets (RockSim); it represents a typical rocket flight from liftoff to apogee...thought it might help you get a sense of the forces involved...

Curious as to your thinking that the GPS velocity may be useful in some way - my understanding of the accuracy of the GPS position would not lend itself to a rocket which is changing its postion so quickly, but you indicate it might do quite well - is the designed-in inaccuracy of the GPS signals their absolute position, but the position change is quite accurate? Accurate enough to track something going 1.5+ mach?

It seems as though, based upon what I have learned so far, and your coments, that none of the typically available sensors are free from the need to reference/sensing gravity except the IR, and maybe the magnetometers?

Could we use a magnetometer for the basic reference? Model rockets typically do not utilize a lot of ferrous material, and other than their time sitting on a big metal launch pad/rail, their environment is pretty metal free...could calibrate away from the pad for the baseline, then once it leaves, it is "free" once again...

How do the "big boys" get around the gravity issue?

Thanks, Frank
At 5:07pm on October 3, 2009, automatik said…
Hi Bill,
I am re-reading your paper "Direction Cosine Matrix IMU" ([pdf] Draft 5/17/2009). I think there is a typo in Eqn. 20 (or there is a chance that I am not getting something). Vector notation, for Z row of the matrix, is represented as follows [rxx, rxy,rxz], and I think that it's a typo as Z row , as I understand it, should be [rzx,rzy,rzz]. What do you think?
At 4:54pm on October 5, 2009, Frank Hermes said…
Hi Bill,

Thanks for the further thoughts...I follow you until you get to the magnetometer and the issue of the single vector availability...so even though you have two or three magnetometers available you still only get one real ref - I guess that is due to the resolution capability of the sensors...

I ran across a guy who told me about a project at Utah State that if you have not seen you might be interested in as it relates to the rocket thing:

If you go to the documentation page, there are some papers descibing their work (see Attitude Control, p.95 in the Flight Readiness paper)...they used a MicroStrain commercial IMU for attitude reference - no GPS...one excerpt from the Navigation section particularly interests me:
A SINS (strap-down INS) works by measuring the accelerations and angular rotation rates of the rocket using an Inertial Measurement Unit (IMU), and integrating these measurements directly to find the velocity and position of the rocket. The IMU measures the sum of all specific surface forces applied to the rocket, including motor thrust and drag, but cannot measure the gravitational force. To account for this, the equations of motion for the rocket
must include the gravitational force before the SINS can integrate and find the rocket velocity and position. SINS is a particularly good measurement device for the thrusting phase of the rocket’s flight, when the motor thrust
varies unpredictably.

A Kalman filter is a mathematically optimal state estimator. It takes multiple measurements, along with their statistical characteristics, and outputs the most probable state of the rocket and its statistical variance. It is unique, in that it is a time-domain filter that requires no system to store previous state data. These qualities make it ideal for inclusion in the navigational algorithm, so that the rocket state is not dependent on a single measurement device.

Sounds to me they just use an estimate for gravity? Thoughts...

Regards, Frank
At 12:49pm on October 6, 2009, Frank Hermes said…
Hi Bill…

The initial goal is to provide a signal (based upon degrees of tilt of rocket tip from vertical) to either inhibit, or possibly trigger, second-stage (sustainer) motor ignition after burnout of the first-stage (booster) motor. The effort is to maximize the coast period between burns. [I attached a flight simulation graph to an earlier email to you to help paint the picture, but I think it may have been reduced to useless by your blog program. If there is a direct email address you can release to me I can resend that graph.]

During booster burn, we expect acceleration along the rocket’s vertical axis to range between 0 and <18 g. After burnout, during coast, deceleration along the rocket’s axis will range from about negative 1.5 g to about 0 g. All other accelerations (both linear and angular) are likely to be quite small since rocket the does not spin much and tips over very slowly.

Booster (first-stage) burn would be between 4 and 12 seconds. The coast period may be from 10-15 seconds. Therefore, the total duration of the sustainer upward flight prior to ignition is less than 15-25 seconds, so indeed long term drift of the sensors may be a minor concern.

In regards to using a magnetometer, the launch rail is typically 10-14’ high and made of steel. I suppose you could calibrate the sensor(s) away from the pad, then mount the rocket, then launch it – once away, there are negligible iron effects. The lull between placing the rocket on the rail and launch could be up to 10 minutes or more.

Further use might include adaptation of the IMU into a vertical-trajectory/vertical-stabilization system to control flight all the way to apogee, which could be as much as one minute flight time.

Regards, Frank

You need to be a member of DIY Drones to add comments!

Join DIY Drones


© 2017   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service