Eh, the only way this will go any farther is if someone gets ahold of proprietary information, so I’m good with resuming the intended topic. If y’all want to dive into the nuances of rotation matrices and Euler angles, I’d be happy to, but as it stands here, this has run its course.
1 Like
If this were of real interest, one could ask Aim or mount a My4 with GPS Module and compare with My5 data. I’ve seen some karts with the My5 unit mounted off the steering wheel. Aim might have an algorithm using the gyros that would allow the MyChron to differentiate between steady state Gs induced by the steering wheel angle and lateral G’s.
I’ve been thinking about this and I agree with @dodo that the kart rotation can be separated from the steering rotation.
The way this can be done is to use the accelerometers to compute a gravity vector and then extract the rotation from the three axis gyro about the gravity vector.
If one wanted, the steering wheel axis can also be estimated because when the kart rotates or the gravity vector rotates, it will be rotating on the steering axis plane.
@HarjitS, to clarify, I said you can correct for both the steering wheel and kart rotations, but you cannot separate them without additional sensors. You can get close using GPS as the additional sensor to estimate kart rotation around a turn, but that’s much coarser resolution than the gyro/accel readings.
You can only determine the gravity vector when static. Dynamically, you can’t assume the orientation of the sensor is the same because both the kart and the steering wheel dance around so much. For example, hitting a curb can cause the whole kart to rotate about the steering wheel plane, so the gyro alone cannot distinguish kart from steering wheel.
I reached out to AIM and asked if the Mychron is compensating for the steering wheel rotation impact to the accelerometer and gps data. Here is their response:
“ Hi Spencer, the Mychron5 does have the internal accelerometer and gyroscope platform. A Mychron5S will not.
Even though the most popular mounting location is the steering wheel, there is not a compensation for steering wheel implications in the accelerometer and gyro scope. Furthermore, you have the GPS calculations of the same measures that do not take into effect these forces to use in comparison.
This video from us touches on Karting analysis and accelerometer readings
There is the ability with connection to RaceStudio3, with the device connected, under Live Measures tab, to calibrate the Accel/Gyro to zero positions in the mounting location to correct mounting angles to better your data precision.”
So TL;DR, they are NOT correcting for the impact of steering wheel angle on the data recorded by the accelerometers and gps accelerometer data.
Interesting to learn that the Mychron5 has accelerometer and gyro platform that the 5s does not have.
4 Likes
Wow, I’m very surprised they dropped the gyro, but I guess that makes sense with the better GPS.
In my defense, I don’t think I said which devices have which sensors, just that gyro → steering wheel correction, haha
Sure but it also demonstrates my point that just because it’s possible, does not mean they are actually doing it. Stating such without confirmation from AIM or access to the source code is just conjecture.
All that to say, I’m glad we got to the truth and confirmed that the lat/long data of both the onboard sensors and GPS data is in fact obfuscated by the changing sensor orientation during corners due to the steering wheel rotation.
Now, can anyone figure out how to correct for this after the fact using a secondary fixed sensor on the kart within race studio or another program? That I’m very curious about.
3 Likes
Very interesting. I found it hard to believe (but it is possible) that the 5S doesn’t have inertial transducers, as the acceleration would have to be calculated by double differentiating the GPS position signal, which would create a very noisy acceleration signal. Although collecting position at 25hz and filtering calculated acceleration signal to 1 hertz might be ok but I would think it would be marginal. Here’s where what was told to Spencer deviates from the Aim website. Aim describes one of their features as “Different Inertial Platform, at 9 axes. This will help with the development of many useful math channels.” Aim Mychron 5S - Aim Technologies
Usually an inertial platform would imply accelerometer and gyros but maybe they are spinning their acceleration and velocity algorithm derived from GPS position as their “Different Inertial Platform”. In addition, I wonder what they mean by “9 axis”, usually there are only 6 axis of motion.
OK I geeked out, sorry.
Well that’s sort of the point of Kp.
1 Like
Often I’ll split the deep dives into separate topics so that the original question is answered more clearly, then if anyone wants to go down the rabbithole they can do so.
9-axis IMU usually means 3 axis accelerometers, gyros and magnetic sensors.
It is surprising to hear that AIM dropped the Accelerometers and Gyros from the Mychron 5S. Maybe they found the data redundant with the more accurate GPS signal. Other than looking at a friction circle, I do not see the benefit of super accurate point to point data. Most of what we use this info for is changes in the curves of the lines on a graph. How deep/hard did I brake? Did I stay within the rubber? Did I get a good exit out of the corner? As long as the source data is consistent, then lap to lap comparisons should be valid.
Personally I am still rocking the Mychron 4 with the chassis mounted GPS09. Unfortunately RS2 does not support Track Map Overlay and RS3 does not support the M4 without juggling a bunch of data around. Despite the more accurate GPS signal on the M5S (down from 3 meters to 1 meter), I have still heard of GPS drift when Track Map Overlay is used. I do not know if AIM has a patch for this yet as I have limited experience with RS3. Last I heard @Ryan_Norberg was still using the M4 with Wheel Speed sensors, because he found it more accurate. Maybe he could chime in on this discussion as far as what data points are more valuable than others. Are Force and Gyro sensors really necessary to improve you driving/setup?
They still have linear accelerometers. If you take the derivative of the GPS speed and compare it to longitudinal acceleration for a 5S, they are basically identical which tells me they aren’t outputting the raw accelerometer data. My guess is they use the accelerometers internally to filter out bad GPS data and then use the accel-informed GPS data for lateral and longitudinal outputs. When done right, this helps fight the drift you get with accelerometers (their big downside) and it helps reduce GPS error. If the GPS is jittery, for instance, you can sanity-check that against a smooth accelerometer reading and know the GPS needs to be numerically calmed down.
I pinged AiM USA about the IMU situation on the MyChron5S and:
On a Kart forum there is a discussion about the Mychron 5s not having accelerometers/gyros (Mychron Gyro, Force sensors Discussion - #24 by dodo) but the product page (Aim Mychron 5S - Aim Technologies ) says it has:
• Different Inertial Platform, at 9 axes. This will help with the development of many useful math channels.
Can you clarify what the Mychron 5S has in terms of accelerometers/gyros?
The response:>
Hello, the Mychron5S versions do not have the accel or gyro internal platform, such as the former, Mychron5.
There are GPS acceleration readings provided in the Data. for an overview of the GPS channels here is a handy FAQ sheet
Your link is to a webpage hosted by the UK distributor, Aim Technologies. They as well as we AiM Sports, the North American distribution group are separate parties from our manufacturing parent of AiM Sportline, Italy.
AiM Technologies has differing web pages; as their own entity chooses. It appears they have likely carried that information over erroneously from the Mychron5 former pages and should correct the error.
The best websites for general AiM Sportline Products and documentation would be found by the official manufacturer site of https://www.aim-sportline.com/ or our US Distribution mirror web page of www.aimsports.com
Here is the link for the Mychron5S technical information from our manufacturing source, let me know if you have any questions.
https://www.aimsports.com/download/technical-sheets/aim_mychron5S_101.pdf
I’m speculating that they compute GPS_gyro by looking at a set of positions to compute a vector and then looking at the rotation of the vector to create GPS_gyro. So, if the unit is rotated about its center i.e. there is no translation (movement), only rotation, it won’t be able to create a rotation output. Again, I don’t know this for a fact but I can’t think of how to get rotation without some translation without a gyro/IMU.
This is of interest to me since I’ve recently been exporting MyChron5S data and doing my own analysis, plotting, etc.
The data which comes out is clearly labeled GPS_ which, to me, indicates it is GPS derived. Since GPS can only provide position and velocity (using Doppler shift) at each time point everything must be calculated from that. As someone else pointed out.
Acceleration - First derivative of speed wrt. time
Bearing - Angle from P1 to P2 between any two consecutive positions
Gyro - First derivative of bearing wrt. time. “Yaw Rate” would be a more accurate description.
From what I can tell they use finite differences to achieve these derivatives. That along with the signal accuracy means by the time you get to the gyro the error could be quite high.
A rigidly mounted 9DoF sensor could easily provide more accurate values at a much higher sampling rate. That $20 part probably cost $800 from AIM…
I think they actually use the accelerometers to adjust GPS position, so when they take the derivative, it is already somewhat sanity-checked by the accels. I’d be very surprised if they use doppler shift for velocity since velocity is perpendicular to the satellite signal. Doppler is most accurate when velocity is aligned to the signal.
Accelerometers, even when rigidly mounted, are not good for slow speed since they suffer from drift. I’ve even seen some pretty bad drift at short, high-speed tests. You can really see this with the 5 (not the 5S) since the 5 outputs the accelerometer data. If you integrate it, you get a nearly monotonically increasing speed trace, so that’s why I think they are differentiating position instead.
GPS position is determined by time using very accurate clocks (accurate to 40 nanoseconds (!!)) on multiple GPS satellites. Your receiver receives time signals from the satellites, compares it to its internal clock then calculates your position. An absolute minimum of four satellites is required to determine your position but it won’t be as accurate as we’re looking for. More satellites is always better but the GPS receiver can only provide your location (X,Y,Z) at a point in time. Everything else (speed, bearing, etc) is calculated by the receiver. The refresh rate of the receiver is what allows the accurate calculation of speed and direction, again the faster the refresh rate the better.
A good explanation of how GPS is used to determine position can be found here: The Global Positioning System: Global Positioning Tutorial.
so motorsport dataloggers like AIM uses doppler shift effect to determine the speed, heading, time. So with these data they can calculate the lat, long G and yaw rate. As long as there are several satellite available, the data is very good, and probably much better than a gyro sensor mounted on a extremely noisy environment like a go kart.
in one of my track cars, i have a AIM MXL dash, which has internal gyro sensor and is fixed mounted in the car. The gyro G channel is extremely noisy, and it was a no brainer to use the GPS G channel instead.
Can confirm the G sensor graph from my Mychron3 Gold was pretty dirty.