Raspberry Pi Pico W LESSON 50: Removing Long Term Steady State Error from Sensor Data

[@nejatpolat]

Great lesson. I love seeing the approach of the engineers to problem-solving. Thanks a lot Paul!

[@paulmcwhorter]

Glad it helped

[@cvicracer]

That channel is awesome i learn alot every time i watch!

[@SeaJay1950]

I appreciate your teaching style and your videos. My New Year's resolution is to stop watching cute cat videos and actually finish one of the 20 or so projects I have started. 😉

[@WhiteError37]

Paul your basically at the point where you've made a lightsaber with the Pico using the mpu6050 and the blade with the neopixel LEDs. Very excited to follow along your teaching is great

[@edSabio572]

Thanks Paul! A belated participation due to illness. US cold weather has not been kind to me. I enjoyed the lesson and working hard to catch up. Note: The concepts taught in the last few lessons have been an eye opener.

[@paulmcwhorter]

I am glad you appreciate them. This is getting into some pretty deep material, but what engineers have to deal with in the real world. Real sensors do not behave perfectly, so you have to use math to get good results.

[@scottpettygrove7821]

That was beautiful! Thank you, Paul and Happy New Year!

[@karl-heinzwaldmann6389]

😊Great, the best food for my brain. It contains everything I need to understand this topic. Reasonable time, appropriate level of difficulty and best practice. Thank you very much

[@crs1548]

Fantastic lesson!

[@lorisrobots]

Great lesson - interesting views of error from an engineering POV. As a statistician, we called these types of errors/variation by different names. I will need to spend some time to connect this to what I know from study of statistics. Awesome stuff!

[@paulmcwhorter]

These are the areas we addressed, using engineering terminology: 1) Noise: A sensor in steady state has random variations about its nominal value. Typically we take the noise out using a low pass filter. 2) Drift: When a sensor is sitting in steady state, its value continues to move in a steady way in one direction. 3) Steady State Error: After applying filters and data fusion techniques, the results have an almost steady long term error. This is driven out by integrating the error, and then removing a proportion of the integrated error each cycle.

[@lorisrobots]

@@paulmcwhorter Great summary - thanks!

[@qzorn4440]

Thanks! Sensational long term error example. Is there a simple Quadcopter educational kit to continue the XYZ sensor programming? Thank you.

[@paulmcwhorter]

Don't know of one. Do really like doing projects like these last few lessons.

[@scottwait3585]

Thank you Paul. Happy new year!

[@arnoldschmucker9977]

This topic was new to me. Kudos!

[@lorisrobots]

Paul - do you have any good references on digital filters like these you have taught us? I would like to go a bit deeper. Thanks for this fantastic series of lessons!

[@paulmcwhorter]

Lori, it is hard to say exactly what I am doing in these lessons, as it is applying/adapting methods from several different area. In some sense, it is Sensor Fusion, as we are combining data from different sensors. That is, the roll and pitch we report is data fused from six different sensors of two different sensor types. Also It we are using a low pass filter, combined with a complimentary filter, or adapted Kalman filter. Then finally, some of the methods draw on the field of PID control. Specifically, when we ended up with the long term steady state error, that we had after all the filtering was applied was removed by integrating the error, and driving the integrated error to zero. This would be an adaption of the "I" math in the "PID" controller. Sorry for the long answer, but I really dont know a single area that formally treats these types of problems. It is adapting and applying know how from different areas.

[@Mircea007]

Hi Paul, we have done something similar with the jetson nano and the object tracking camera. on a project of my own I overcomplicated the code with lots of if statements to filter the signal from a throttle pedal and for example when the driver studently comes off the pedal the code will not filter the signal and the engine speed comes down without delay.

[@albrody8861]

Great lesson and no we don't hate you!

[@jamescullins2709]

Outstanding Paul! thanks

[@pgranzow5677]

Hey Paul, I started with Arduino a few weeks ago and I watched your new improved Arduino Tutorials... I love them, but I slowly reach the end of your Playlist. So the question is, will you do more Arduino Tutorials, after Rasberry Pi? Because I really learned a much in these Vids. And if there are more, what are they about? Keep the amazing Videos going... greetings from Germany.

[@paulmcwhorter]

I am going to make a new series on the Arduino R4. Should start soon.

[@charlotteswift]

@pgranzow5677 Unfortunately, I believe that Paul is going to start from the beginning again.😿I would prefer it if he suggested that absolute beginners watch the previous series of lessons before the new series and therefore the R4 lessons could be more useful. I know that many existing subscribers have said that they won't be watching.

[@DrDave327]

Even with reviews of basic concepts, Paul has, in the past, slipped in a few context items that challenges the experienced students. Also, solving the basic homework really reinforces the mastery of the concepts.@@charlotteswift

[@blitzlures4563]

Very interesting. Thank you.

[@paulmcwhorter]

Glad you enjoyed it

[@alistaircook1997]

enjoying your videos,I was looking at the arduino R4 wifi, and found any library that controls the arduino R3 Neopixel strip WS2812, does not work for the R4, do you know of a way to get the R4 to light the Neopixel strip ?.

[@noweare1]

Hello Paul, noticed that the title is missing the W. It is in all the other titles and I know you like to be precise.

[@taurus548]

WOW!

[@DrDave327]

Nice exercise, Paul! Tough assignment. I believe this tweaked version is functioning better: Now the OLED seems to be reporting the angle in degrees. The bubble seems to be functioning best with only the accelerometer data. Thanks for the challenging assignment! Here is the video: kzbin.info/www/bejne/rmWci6xumrJlitk

[@paulmcwhorter]

LEGEND!

[@jameslewellen150]

Here is the homework for this lesson. Nice lesson, I had trouble doing this homework for the previous lesson. This video cleared it up nicely. Here is the link to my video: kzbin.info/www/bejne/fpa2g5mjo717oZY

[@paulmcwhorter]

Excellent work. Dont feel bad you did not do it on your own. This would be Master's Level Electrical Engineering course material. It is good you were able to understand it and do it once I showed. Well done!

[@briankelly4095]

Hi Paul. Thanks for another great lesson. You'll find my Lesson 50 homework here kzbin.info/www/bejne/j4O5e5-clL-smsU

[@paulmcwhorter]

LEGEND!