IMU data viewer? Yes, with MatLab.

[RichWest]

I'm just looking for a graphical viewer (chart) for some IMU data I'd like to display visually. The data is from the Variense VMU931, recorded or streamed 9DOF of data. I'll admit I'm being lazy, but my interests is in displaying the data rather than learn another program. Still designing a test fixture, but that is getting closer. The VMU931 is a great size/weight and the enclosure makes installation a breeze.

The Variense Reader is fairly basic visually, I'm mean really basic... It displays the live sensor stream and records.
I looked at MatLab, I'm not a programmer nor an engineer.... This seems to be the default application for engr'ing types. But then it appears intimidating for the non-programmers, like myself.
Excel is just not something I've enjoyed using, never works well with me, at least when making charts.
Blackbox Explorer, at least how it's shown to be, seems fairly dynamic. Not sure it can import CSV files and display.
Mission Planner, haven't figured how to import a CSV file...

Any help or suggestions would be appreciated. Just want to move on with my prototype testing without having to work out learning to program software...

 

[RichWest]

Like a needle in a haystack, but I did find a couple of great resources for vibration and data interpretation. If the subject is of interest to you, then these will give you some direction and answers.

Tom Irvine has both MatLab and Python scripts for displaying csv data. He's a well respected engineer in the community of engineering types...rocket science and such. If you are interested in mathematics and vibes, this is The Guy. I'll be using his MatLab scripts/librarys for my data, as he has already done the math...

Vibrationdata

Then Mide Technology, maker of the Slam Stick, has several great videos and white papers on vibration discussion.

Mide Technology | Engineering Solutions & Products

Steve Hanly, with Mide Technology, produced an overview video regarding vibration testing. It's an hour long video that is not overly technical, but covers most of the considerations needed when doing this type of testing. He presents in a DIY and/or open source thinking, not a true marketing video as initially expected. Well worth a watch if of interest....

Shock and Vibration Testing Overview

 

[RichWest]

Here is a video showing a pair of lasers vibrating from my test fixture and simulated gimbal. I just wanted a visual baseline before I start hooking up the sensors. Building the fixture has been a challenge, its taken 8 weeks to get to this point. But then the results are promising, so I keep moving forward.

I always thought Houston was huge and had everything a person could need local. Not so any more, had to order most of the material online. Couldn't find a proper 2.5mm screw locally...this town sucks...

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[RichWest]

I'm on a big learning curve with all this, high level math was not my subject. Not sure I'm on the right path for plotting the data, so this is just for the sake of conversation....if of interest to others.

Still testing my fixture and sensors for consistency of results. Excel is my default for plotting the data stream until I learn more about the subject. For plotting it seems the Gyro Y & Z data plots well enough to see what is going on. Accel data is equally interesting, but it is just sine waves of amplitude and frequency. Again, learning by doing.

Here is a couple interesting examples to compare what I'm seeing so far, it's about 0.5 sec. of data captured at 1kHz. As I increase the number of data points in the graph it becomes real obvious that the stock dampeners are really dirty vibe wise, the trendline being the visual reference.

If you will, this is your camera movement while in (simulated) flight off a static gimbal. Kind of an extreme case as an actual 3 axis gimbal would further reduce the spread seen, especially in the Y axis. Plus I'm intentionally driving up the vibes with unbalanced motors for the purpose of revealing the dampening systems' flaws. I'm able to generate frequencies up to 200Hz comfortably, beyond that the dog and children run for cover...

 

[RichWest]

I wanted to followup with my adventures to plot vibration data from a camera's perspective.

Too much programming for MatLab, so I stuck with Excel and found a reasonable method to display the data. Someday MatLab, but time better spent elsewhere at the moment...

I somewhat agree with Elon Musk, “Anyone relying on lidar is doomed. Doomed! ". Computer vision is the path forward for many autonomous vehicles.

Below I've charted some example accel data I've collected. The plots are of the vibration input (plate) and the damped results (gimbal). I've overlaid the Plate and Gimbal in a chart to give some comparison between the two sets of data. XY and YZ charts display two perspectives. XY is a gimbal top down look and YZ is from the camera lens/sensor's perspective on the gimbal. My tests are typically an exaggerated amount of vibration amplitude, well beyond what a typical drone should be producing. My purpose is to excite the dampers to an exaggerated level for the plotted data to reveal a pattern. As it is, frequency doesn't change much with the levels of amplitudes I'm dealing with.

This 0.25 sec. of accel data was recorded at 1kHz, to put it in perspective that is 7 frames of video at 30fps, The resulting vibration/oscillation, even when amplitudes are scaled down, is effectively causing a frame level blur. In the CV world they refer to it more as pixel blur and blur beyond 4 pixels is consider marginal to identify objects consistently. You photogs see this pixel issue as soft, out of focus or blown out highlights in your images. You compensate this with ND filters and clean up in post, as best you can. And as well you choose the "ideal" light and angles. CV doesn't have that luxury of time, light or good angles, they need to process images at a fairly high rate to analyze the 3D world or count chickens on the fly.

Finally, the observed oscillation (or similar) below does affect pretty much all UAVs with vision devices, gimbaled or otherwise. Just the nature of the beast. My interest is to dynamically mitigate as much vibration as is mechanically possible....which appears to be an endless rabbit hole to a critical damped solution. I find the topics challenging and interesting. How do you eat an elephant...

 

[MRhoades]

A licensed copy of Matlab is very expensive. I am engineer and use it nearly every day for my job.
An alternative is Octave. Octave is an open source free application that is around 95% compatible with MATLAB code.
It can run most Matab code and MATLAB can run code created in Octave.

You may want to check it out.


- Mike

 

[RichWest]

A licensed copy of Matlab is very expensive. I am engineer and use it nearly every day for my job.
An alternative is Octave. Octave is an open source free application that is around 95% compatible with MATLAB code.
It can run most Matab code and MATLAB can run code created in Octave.

You may want to check it out.


- Mike

Way cool, had no clue there was an alternative program with MatLab capabilities... There is a limited cheaper version of MatLab for home use for $3-400 a year. Learning Python is the other hurdle for either case, just don't have the will to learn another language at this point in life...

Linked above is Vibration Data's website, with many Python/MatLab scripts, so I'll be able to give those a try now. Thanks again for the suggestion of Octave.