Data Acquisition in Sx/Mx

Motodave15
Posts
4252
Joined
8/7/2010
Location
Temple City, CA US
Edited Date/Time 1/22/2022 1:30pm
Ok so I have a Information Technology background and am wondering how I could stop dealing with the regular corporate people and transition my I.t. Skills into Racing industry (racing is my passion period). (Not limited to motocross either, I’d work for a nascar/Indycar/rally/f1 team in like .01 of a second)

Do any of the teams run Data Acquisition (Telemetry or otherwise) when practicing and need to use a cloud to hold the data and or getting that configured or live time data transmission of such things? If anybody has any insight into this it would be appreciated… side note, I’m going to start helping a off-road race team and I’m hoping to find a way to bring I.t data into the fold to learn more/progress.
2
1
|
ML512
Posts
16857
Joined
12/28/2008
Location
Wildomar, CA US
Fantasy
1/20/2022 6:01pm
Motodave15 wrote:
Ok so I have a Information Technology background and am wondering how I could stop dealing with the regular corporate people and transition my I.t. Skills...
Ok so I have a Information Technology background and am wondering how I could stop dealing with the regular corporate people and transition my I.t. Skills into Racing industry (racing is my passion period). (Not limited to motocross either, I’d work for a nascar/Indycar/rally/f1 team in like .01 of a second)

Do any of the teams run Data Acquisition (Telemetry or otherwise) when practicing and need to use a cloud to hold the data and or getting that configured or live time data transmission of such things? If anybody has any insight into this it would be appreciated… side note, I’m going to start helping a off-road race team and I’m hoping to find a way to bring I.t data into the fold to learn more/progress.
A few of the teams run data acquisition during certain test sessions and occasionally at the races. Some teams more than others.

I think the amount that use it doesn't create enough need for services that would keep someone busy as a job.

I may have miss understood the later question, but the teams can't receive or send a live signal to the bikes in race trim.
1
Motodave15
Posts
4252
Joined
8/7/2010
Location
Temple City, CA US
1/20/2022 6:06pm
ML512 wrote:
A few of the teams run data acquisition during certain test sessions and occasionally at the races. Some teams more than others. I think the amount...
A few of the teams run data acquisition during certain test sessions and occasionally at the races. Some teams more than others.

I think the amount that use it doesn't create enough need for services that would keep someone busy as a job.

I may have miss understood the later question, but the teams can't receive or send a live signal to the bikes in race trim.
Ahh interesting ok, yeah you answered my question completely.
ML512
Posts
16857
Joined
12/28/2008
Location
Wildomar, CA US
Fantasy
1/20/2022 6:11pm
Motodave15 wrote:
Ahh interesting ok, yeah you answered my question completely.
By example, when LITPro first became a big deal, they had a staff member or two go to the races and help track the data, store it, do reports, and display it for the teams. When it came down to the teams paying for their services, only KTM really kept their guys around...and that only lasted another year or two.
4
sandtrack315
Posts
2823
Joined
7/19/2013
Location
Philadelphia, PA US
1/20/2022 6:18pm Edited Date/Time 1/20/2022 6:20pm
Check out Ducati, they have hired some data scientists. You’ll have to know machine learning, though.

https://analyticsindiamag.com/ducati-goes-big-data-machine-learning-imp…

I have thought about going this way, but I think the way to get paid less than working at a university is working on a race team…
7

The Shop

mxtech1
Posts
1968
Joined
7/21/2011
Location
Galesburg, IL US
1/20/2022 6:30pm
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
20
1
Bruce372
Posts
6329
Joined
4/1/2008
Location
US
1/20/2022 6:32pm
Check out Ducati, they have hired some data scientists. You’ll have to know machine learning, though. https://analyticsindiamag.com/ducati-goes-big-data-machine-learning-improve-motogp-bikes-testing-process/ I have thought about going this way, but I...
Check out Ducati, they have hired some data scientists. You’ll have to know machine learning, though.

https://analyticsindiamag.com/ducati-goes-big-data-machine-learning-imp…

I have thought about going this way, but I think the way to get paid less than working at a university is working on a race team…
If you can do machine learning, get into biotech. You might help create the next Amgen or Genentech
1
1
mxtech1
Posts
1968
Joined
7/21/2011
Location
Galesburg, IL US
1/20/2022 6:33pm
ML512 wrote:
By example, when LITPro first became a big deal, they had a staff member or two go to the races and help track the data, store...
By example, when LITPro first became a big deal, they had a staff member or two go to the races and help track the data, store it, do reports, and display it for the teams. When it came down to the teams paying for their services, only KTM really kept their guys around...and that only lasted another year or two.
I would argue that KTM figured out how to use their own GPS sensor, tied directly to the factory ECU, and then combined it with ECU CAN data, which would provide for superior data collection system than what the LITPro guys could offer with only time-series based GPS data.
1
brocster
Posts
4472
Joined
6/9/2009
Location
Aliso Viejo, CA US
1/20/2022 6:53pm
mxtech1 wrote:
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how...
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
I’ll play. We have been using those string pot measuring devices on control valves in the refining/petrochemical industry for years (2000 ish before smart positioners) to do signatures and diagnostics on critical control valves for complex control systems. Looking at step test reveals a lot about valve performance.

With all of that said and you guys mentioning live on track data via a cloud or even a wireless hub perked my ears on a possibility of an overlay system that could possibly be very beneficial. MXDave15 and mxtech we maybe should talk. I am not an engineer nor a IT person but have to dabble a little in each enough to be dangerous.
2
1/20/2022 6:57pm
LitPro’s limitations centralize around the disconnect between the rider and bike. Knowing lap times, Gs, jump height, etc. is all really cool. I have one and like playing with it. Beyond that it leaves you wanting for something that links to bike telemetry to let you really analyze technique and bike behavior. Knowing throttle position, gear, engine rpm, wheel speed, etc would add a ton of value, similar to what mxtech mentioned with the KTM ecu. This would even create additional product lines if they provided sensors. I bought the product originally because I thought this is the direction the company was going in, but that hasn’t been the case.
ML512
Posts
16857
Joined
12/28/2008
Location
Wildomar, CA US
Fantasy
1/20/2022 7:02pm Edited Date/Time 1/20/2022 7:07pm
ML512 wrote:
By example, when LITPro first became a big deal, they had a staff member or two go to the races and help track the data, store...
By example, when LITPro first became a big deal, they had a staff member or two go to the races and help track the data, store it, do reports, and display it for the teams. When it came down to the teams paying for their services, only KTM really kept their guys around...and that only lasted another year or two.
mxtech1 wrote:
I would argue that KTM figured out how to use their own GPS sensor, tied directly to the factory ECU, and then combined it with ECU...
I would argue that KTM figured out how to use their own GPS sensor, tied directly to the factory ECU, and then combined it with ECU CAN data, which would provide for superior data collection system than what the LITPro guys could offer with only time-series based GPS data.
LITPro helped them develop the onboard data system they're using now. They sorta tied in the existing data from the bike they were already gathering, such as gear position, throttle position, and wheel speed...then tied it to suspension data and I believed it's laid out in an easier to digest format more similar to their consumer product.

I probably should've mentioned that in the above post that KTM did continue with them, just in a different format.

Outside of the major race teams, the standard wiring harnesses on these bikes are going to need a major overhaul/upgrade for any regular consumers to utilize data systems.
1
ML512
Posts
16857
Joined
12/28/2008
Location
Wildomar, CA US
Fantasy
1/20/2022 7:05pm
Don't get me wrong, I'm not suggesting that better acquisition systems wouldn't bee beneficial. My comments were more geared towards not seeing enough marketplace for there to be a financially viable product.

2D and Aim are the two primary systems I see most teams using. Most of them using 2D.
KurtJ99
Posts
2354
Joined
2/6/2017
Location
CA US
1/20/2022 7:20pm
One of the challenges in collecting data is deciding what to do with it. In road racing there are less dimensions to analyze with a flat surface and traction conditions that are bounded. MX/SX tracks have changing track conditions both on different areas of the track and over time. Do you want to steer or track with the rear wheel? Can you infer the optimum tire pressure by understanding g forces, lean, and power transmitted? Or does your data tell you the obvious and 12 psi is about the same as 12.5? Tough to say but you can’t find out if it’s a differentiator unless you invest in the tech long term.
3
Grouper
Posts
300
Joined
11/28/2021
Location
Paducah, KY US
1/20/2022 7:29pm
mxtech1 wrote:
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how...
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
I didn’t read all of that I’m in a movie but I’ll say that that’s really true. I’m a mechanical engineer and most people would be shocked at how little many people in industry know little outside of it. Anyways, I’ve been thinking a lot about stuff like this too. If either you or OP want to chat more let me know. 5616170259
1
MxKing809
Posts
6374
Joined
10/13/2013
Location
Big Sand Whoops, MI US
Fantasy
1/20/2022 7:43pm
Man…. Maybe we need to start a discord channel or group chat about this topic for further conversations.

Operation manager here in the utility space, with expertise in mobile field applications, GIS, ioT, algorithm writing and data management.

What’s a Vortex ECU run? A grand? I’m pretty sure someone could market a system like the Aim platform with a mobile app and UI that brought in telemetry from a la carte sensors that were modular integrations into the ECU and data lake.
1
mxtech1
Posts
1968
Joined
7/21/2011
Location
Galesburg, IL US
1/20/2022 7:48pm Edited Date/Time 1/20/2022 7:50pm
LitPro’s limitations centralize around the disconnect between the rider and bike. Knowing lap times, Gs, jump height, etc. is all really cool. I have one and...
LitPro’s limitations centralize around the disconnect between the rider and bike. Knowing lap times, Gs, jump height, etc. is all really cool. I have one and like playing with it. Beyond that it leaves you wanting for something that links to bike telemetry to let you really analyze technique and bike behavior. Knowing throttle position, gear, engine rpm, wheel speed, etc would add a ton of value, similar to what mxtech mentioned with the KTM ecu. This would even create additional product lines if they provided sensors. I bought the product originally because I thought this is the direction the company was going in, but that hasn’t been the case.
The system you’re describing is already offered by AiM.

Taipan ECU
-> AiM SmartyCam + an AiM GPS08 module
-> AiM data hub
-> ECU CAN connector
-> AiM memory module

Download data after each ride then can use AiM software, or something much more capable, like nCode, to post-process. This would give you the ability to create videos, with time matches CAN data like throttle percentage, gear selection, engine RPM, etc.

You can load up additional sensors (gyro, string pots, rotational, pressure, etc,) by adding additional multi-port AiM data hubs. No big deal.

LitPro has a great product, but they focus solely on the software side of data. They stopped investing in hardware, which may limit them in the future, unless they figure out how to partner with 3rd party electronic manufactures to tie the ECU CAN traffic into their suite.
2
mxtech1
Posts
1968
Joined
7/21/2011
Location
Galesburg, IL US
1/20/2022 7:52pm
KurtJ99 wrote:
One of the challenges in collecting data is deciding what to do with it. In road racing there are less dimensions to analyze with a flat...
One of the challenges in collecting data is deciding what to do with it. In road racing there are less dimensions to analyze with a flat surface and traction conditions that are bounded. MX/SX tracks have changing track conditions both on different areas of the track and over time. Do you want to steer or track with the rear wheel? Can you infer the optimum tire pressure by understanding g forces, lean, and power transmitted? Or does your data tell you the obvious and 12 psi is about the same as 12.5? Tough to say but you can’t find out if it’s a differentiator unless you invest in the tech long term.
I’d rather have data and have to decide what to do with it than have no data and make a decision with less information.

That’s a No brainer hombre.
1
sandtrack315
Posts
2823
Joined
7/19/2013
Location
Philadelphia, PA US
1/21/2022 4:20am
Check out Ducati, they have hired some data scientists. You’ll have to know machine learning, though. https://analyticsindiamag.com/ducati-goes-big-data-machine-learning-improve-motogp-bikes-testing-process/ I have thought about going this way, but I...
Check out Ducati, they have hired some data scientists. You’ll have to know machine learning, though.

https://analyticsindiamag.com/ducati-goes-big-data-machine-learning-imp…

I have thought about going this way, but I think the way to get paid less than working at a university is working on a race team…
Bruce372 wrote:
If you can do machine learning, get into biotech. You might help create the next Amgen or Genentech
I work in biotech at UPenn. I’m a senior data scientist in a neuroscience lab. I handle all the informatics, machine learning, coding architecture, and publication workflows in the lab along with two other senior scientists, one of whom got poached to work at J&J and is leaving in a week ha.
1/21/2022 4:38am
Motodave15 wrote:
Ok so I have a Information Technology background and am wondering how I could stop dealing with the regular corporate people and transition my I.t. Skills...
Ok so I have a Information Technology background and am wondering how I could stop dealing with the regular corporate people and transition my I.t. Skills into Racing industry (racing is my passion period). (Not limited to motocross either, I’d work for a nascar/Indycar/rally/f1 team in like .01 of a second)

Do any of the teams run Data Acquisition (Telemetry or otherwise) when practicing and need to use a cloud to hold the data and or getting that configured or live time data transmission of such things? If anybody has any insight into this it would be appreciated… side note, I’m going to start helping a off-road race team and I’m hoping to find a way to bring I.t data into the fold to learn more/progress.
Where I work we develop custom measurement and acquisition systems, most of our work comes from places/clients where off the shelf products don't fit their needs. I think your best bet would be to broaden your targeted industries, still push to get into the mx industry but also target motorsport, transport, military etc. The common ground here is that the system and sensors need to be small, light and withstand some pretty horrible conditions.

You will also need to have some good mechanical design, fabrication and machining skills because in my experience, a huge part of the challenge for this work is being able to retro fit sensors in a way that doesn't damage or affect the performance of the system you are trying to measure. If you can be a one stop shop you have a better chance of being able to bring the cost down to a point where you can attract customers, but also get reasonably compensated for your time. If you end up having to outsource design and machining on top of sensors etc. it becomes infeasible very quickly.

Just my two cents. Anyway cheers for making an interesting thread hope you can find a way to align your work with your passion!
3
forensic
Posts
117
Joined
12/3/2011
Location
Madison, WI US
1/21/2022 5:14am
mxtech1 wrote:
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how...
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains mostly stationary to the bike and the road surface is smooth and uniform. In those cases you can make good sense of logged data with just some on-board video to help with context. Motocross is difficult because of the rider-active nature, large movements of the rider's c.g. and large variances in forces on the handlebars and footpegs. To make sense of logged data you need to have lots of video and lots of measurements of the track surface, and then special software to combine the three data sources (logged data, track measurements, and external video).
2
sandtrack315
Posts
2823
Joined
7/19/2013
Location
Philadelphia, PA US
1/21/2022 6:38am
mxtech1 wrote:
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how...
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
forensic wrote:
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains...
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains mostly stationary to the bike and the road surface is smooth and uniform. In those cases you can make good sense of logged data with just some on-board video to help with context. Motocross is difficult because of the rider-active nature, large movements of the rider's c.g. and large variances in forces on the handlebars and footpegs. To make sense of logged data you need to have lots of video and lots of measurements of the track surface, and then special software to combine the three data sources (logged data, track measurements, and external video).
Yeah, one of the main issues for machine learning in MX is variance in the data relative to the amount of data. Say I am collecting data on how the fork and shock work with difference linkages, and how this impacts section times. It's going to take a ton of laps to get enough data to differentiate linkages, as the variance lap to lap is going to be pretty high. Plus, as the day goes on, the track changes. I am not saying it's not possible, but typically I work with millions of data points in order to build an accurate model, and there is no where near the structured variance (e.g., track gets rougher throughout the day).

One day there will certainly be a data scientist at factory Honda, but they are not going to get someone good enough to make progress unless they are paying him/her at least 250k. Correct me if I am wrong, but that would probably be a lot more than the team manager.
1
Richy
Posts
3086
Joined
7/18/2020
Location
UK GB
1/21/2022 6:43am Edited Date/Time 1/21/2022 6:44am
I'll be honest I work with ECU's, dashes, data loggers and the remainder of the electronics side, albeit mostly wiring and assembly nowadays and almost exclusively racecars.

I don't really understand why normal people want to see more tech in Motocross. As in not the teams hunting titles.

There are tonnes of high tech, mega money, mm motorsports and rarely does it lead to better racing and more fun for the majority of people involved.

I appreciate electric start as much as the next guy, sometimes, but the (relative) simplicity of a motocross bike is one of the things that makes them so cool in my humble opinion.
8
ktmdan
Posts
1226
Joined
4/1/2008
Location
Houston, TX US
1/21/2022 6:48am
mxtech1 wrote:
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how...
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
forensic wrote:
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains...
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains mostly stationary to the bike and the road surface is smooth and uniform. In those cases you can make good sense of logged data with just some on-board video to help with context. Motocross is difficult because of the rider-active nature, large movements of the rider's c.g. and large variances in forces on the handlebars and footpegs. To make sense of logged data you need to have lots of video and lots of measurements of the track surface, and then special software to combine the three data sources (logged data, track measurements, and external video).
I suppose you could put load cells in the bars, pegs, and a few in the seat. Combine that with some markerless motion capture and you could come up with some useful data.

The problem is this would be overly complicated and extremely expensive. And at the end of the day does it really matter??? It's just dirt bikes after all...
1/21/2022 8:52am
I always thought it would be interesting to see an overlay of throttle position and rpm as a bike skims over the whoops. Assuming a 125 or 250F is WFO would like to see how revs move up and down as bike blitzes the whoops.
1
slowgti
Posts
1005
Joined
1/14/2016
Location
Monroe, GA US
1/21/2022 1:44pm
Richy wrote:
I'll be honest I work with ECU's, dashes, data loggers and the remainder of the electronics side, albeit mostly wiring and assembly nowadays and almost exclusively...
I'll be honest I work with ECU's, dashes, data loggers and the remainder of the electronics side, albeit mostly wiring and assembly nowadays and almost exclusively racecars.

I don't really understand why normal people want to see more tech in Motocross. As in not the teams hunting titles.

There are tonnes of high tech, mega money, mm motorsports and rarely does it lead to better racing and more fun for the majority of people involved.

I appreciate electric start as much as the next guy, sometimes, but the (relative) simplicity of a motocross bike is one of the things that makes them so cool in my humble opinion.
As a fellow dater tater, I agree. I think you would have better luck in imsa/sro/wec finding a job.
1
forensic
Posts
117
Joined
12/3/2011
Location
Madison, WI US
1/21/2022 1:57pm
mxtech1 wrote:
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how...
In my opinion, Factory, and even satellite level teams, are years behind the curve in terms of what can be done with data acquisition and how interpretation of said data can be used to drive engineering decisions. Part of the reason why, is very few of the crew members, managers, and mechanics have exposure outside of the motocross industry as to what can actually be done with data analytics and how powerful it is. When you have an industry that tends to promote and hire from within the “bro” pool, you tend to hold onto tried and tried methods fro developing race bikes. If you take a sample of most major OEMs (auto, off-highway, power products, etc) every other industry is using data to drive business and in this day in age Big Data is just as important as hard iron, but I’m preaching to the choir if you come from an IT background.

I work as a test engineer and on a daily basis, deal with collecting, analyzing, and interpreting data. In our industry, it’s frowned upon to “engineer from the gut” meaning you better have data to support your decision. This mentality doesn’t exist in moto. It’s all by feel and rider to crew interpretation of what’s happening on the track. I truly feel the lack of data to drive decisions is why historically certain riders and teams are constantly chasing their tails throughout the season, looking for the magic combination of parts and setup. The bikes and riders are so competitive today, that we are getting to the point where splitting hairs on % setup improvement will be done with data and engineering, not what a rider “thinks” they feel. Not saying the rider element will ever be eliminated, but data can and should play a much bigger role.

I remember a story on pulp where KYB had collected some on-track suspension data. They brought over Honda CRF engineers, who were good engineers, but had NO IDEA how to interpret the on-track and what it meant. From the story, they spent a significant amount of time working with the rider and the KYB tech to gain a grip on what the data actually meant. Once they understood how to interpret the suspension data, they started making massive gains in performance. Proof that data enables hidden performance, as this example was Tomac’s final year on the Geico Honda 250 team where he absolutely murdered everyone on that bikes.

I see soo much potential for data acquisition in MX/SX. 32-bit ECUs, with input/output channels and CAN terminals, like the AiM Taipan, is a massive step-function in the right direction to simplify a data logging system. For the life of me, I can’t understand why the major Factory teams don’t employ some sort of a data analysis/mechanical engineer that is viewing every bit of data when the bikes come off the track. This would be a major advantage over time. When you transition from making decisions because of data, instead of off the cuff, you start getting millions and millions of data points, then you start modeling AI and ML models that can can trained to simulate the outcome of a change before a part is even swapped or a tuning adjust is made. Think about hours many hours of physical testing that would save. F1 has laid the groundwork for what’s possible, even if SX/MX would follow that model to even a fraction of a percentage, it would start to revolutionize how the teams are managed.

If you ever want to talk offline about starting some sort of 3rd party data service company, specifically targeted at MX/SX, hit me up. I could talk about this topic for days!!
forensic wrote:
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains...
I'm a mechanical engineer currently working on a personal project just like this. I've done a lot of testing on street bikes where the rider remains mostly stationary to the bike and the road surface is smooth and uniform. In those cases you can make good sense of logged data with just some on-board video to help with context. Motocross is difficult because of the rider-active nature, large movements of the rider's c.g. and large variances in forces on the handlebars and footpegs. To make sense of logged data you need to have lots of video and lots of measurements of the track surface, and then special software to combine the three data sources (logged data, track measurements, and external video).
ktmdan wrote:
I suppose you could put load cells in the bars, pegs, and a few in the seat. Combine that with some [url=https://www.swri.org/industry/biomechanics-human-performance/markerless-motion-capture-laboratory]markerless motion capture[/url] and you...
I suppose you could put load cells in the bars, pegs, and a few in the seat. Combine that with some markerless motion capture and you could come up with some useful data.

The problem is this would be overly complicated and extremely expensive. And at the end of the day does it really matter??? It's just dirt bikes after all...
you are right on. It's very expensive equipment but fortunately I have it. I scanned a track feature with a terrestrial scanner as well as the position of a camera. Using VFX software I pixel tracked the bike and rider through the scene, and synched it with an onboard data logger. Strain gauges on the handlebars in addition to the usual IMU and sensors. Does it matter? I enjoy it and learning a quite a bit.
2
brocster
Posts
4472
Joined
6/9/2009
Location
Aliso Viejo, CA US
1/21/2022 3:51pm
My thinking is that with the rider have to adapt to changing track conditions with body position and speed of reaction and with those being endlessly variable the only things that could improve with data capture is bike set up. i.e. suspension and traction. With all other points being inconsistently variable I an not sure what its worth just for collection and analysis. Tuning for situation “x” when there are no guarantees that it can be duplicated in the same spot lap after lap. Interesting as hell though.
Motodave15
Posts
4252
Joined
8/7/2010
Location
Temple City, CA US
1/21/2022 4:09pm
forensic wrote:
you are right on. It's very expensive equipment but fortunately I have it. I scanned a track feature with a terrestrial scanner as well as the...
you are right on. It's very expensive equipment but fortunately I have it. I scanned a track feature with a terrestrial scanner as well as the position of a camera. Using VFX software I pixel tracked the bike and rider through the scene, and synched it with an onboard data logger. Strain gauges on the handlebars in addition to the usual IMU and sensors. Does it matter? I enjoy it and learning a quite a bit.
That’s interesting right there. I’d like to see some of that in action if you have any demos or YouTube of it.
2
TrailMaker
Posts
104
Joined
1/9/2022
Location
Fresno, CA US
1/21/2022 4:41pm Edited Date/Time 1/21/2022 4:54pm
Data in MX often does not translate to what the rider feels, and how the bike rides. A good example of this was when Yamaha started switching over from steel frames to aluminum frames manufactured from castings and forgings that were then welded and CNC machined. The design engineers had enough on their plate with the manufacturing issues, they they designed the new bikes with the exact same kinematic geometry, and the frame design was tweaked to match the stiffness and elastic characteristics of the prior bikes. They did lots of FEA analysis, and load testing on prototype frames till they got it "right". The geometry of the overall bike/frame was the same, and the stiffness of the frame overall, and in localized areas matched the prior bike. "Exactly the same in every measure" the Engineers insisted. Every rider said the Al bikes felt NOTHING like the prior years bike they were trying to mimic. Go figure....

Set up and design in MX is more about "translation" of what the rider feels to engineering principles than it is about "measuring" things and drawing conclusions without filtering through the rider. Also, if you think that the mfgs do not have state of the art data engineering, then you are delusional. They have access to equipment we have probably never heard of. If they are not regularly using it, it is because of the rider vs measurement usefulness issue.
4
1
kibby
Posts
84
Joined
1/9/2012
Location
El Cerrito, CA US
1/21/2022 8:43pm
I always thought it would be interesting to see an overlay of throttle position and rpm as a bike skims over the whoops. Assuming a 125...
I always thought it would be interesting to see an overlay of throttle position and rpm as a bike skims over the whoops. Assuming a 125 or 250F is WFO would like to see how revs move up and down as bike blitzes the whoops.

3
kibby
Posts
84
Joined
1/9/2012
Location
El Cerrito, CA US
1/21/2022 8:55pm
I always thought it would be interesting to see an overlay of throttle position and rpm as a bike skims over the whoops. Assuming a 125...
I always thought it would be interesting to see an overlay of throttle position and rpm as a bike skims over the whoops. Assuming a 125 or 250F is WFO would like to see how revs move up and down as bike blitzes the whoops.
kibby wrote:
[img]https://p.vitalmx.com/photos/forums/2022/01/21/529723/s1200_Screenshot_158.jpg[/img]


2

Post a reply to: Data Acquisition in Sx/Mx

The Latest