PSA: Inmotion V13 alarms and tiltback combined design flaw makes it unsafe to ride at high speed. Inmotion working on a fix

[supercurio]

The rider Meriku experienced an overlean crash on his V13 today.
Source on Telegram: https://t.me/inmotionV13eng/7026 and Facebook: https://www.facebook.com/groups/inmotioneuc/permalink/3036997299928559/

Here's the video he captured of it:

 

What happened?

1. The wheel alerted the rider when reaching 72.1 km/h during strong initial acceleration
2. The rider eased up on his forward lean and acceleration following the alert
3. Then resumed accelerating since the wheel stopped beeping anymore.
4. When reaching 81.2 km/h, the wheel overleans. No beep, no tiltback but a crash instead.

Inmotion's response

Inmotion is currently analyzing the logs they obtained from the rider.
They acknowledged that there is not consistent beeping and will find the root cause. (source: CEO Bob in a private discussion)

Edit: In a Facebook discussion, Mark Yu, a Inmotion employee (YouTube channel) unofficially confirmed the root cause which is the same as what I identified and reported 2 years ago on the V10F.

Update 17 February:
Inmotion published a statement:
https://www.facebook.com/groups/inmotioneuc/permalink/3037833299844959/

 

But I thought that the V13 couldn't cutout!

With current technology limitations, every self-balancing vehicle can be overleaned with sufficient combination of speed and acceleration demand from the rider.

This shows what happens when ignoring the beeps and cutting out as a result, like we've seen many times before.
What is different here however is that who ignored the beeps was not the rider but the wheel itself instead.

Are there previous cases?

It is not the first time that Inmotion wheels don't beep when they are supposed to. Both the V10F and V11 have been affected by such issue, which was supposed to be fixed on the V11 after a crash. The V10F is still affected. I reported the issue 2 years ago here:

The reply from Mark Yu on Facebook (product manager at Inmotion) develops further on that, reading:
sources: 1 2 3 4

Quote

Inmotion wheels alarm beeps every 5 seconds. That's too long.

François Simond Not bug. The software control the over power alarm sound beeps every 5 seconds. It's a shit design but not a bug
François Simond The speed alarm beeps is much shorter. But others beeps just have 5 seconds interval. Bad design

Update February 17:

Another rider experienced the same 5s alarm delay and reached the wheel speed of 58mph, luckily didn't crash but must have been very close to. Seen here with timestamp.

 

What about tiltback?

Tiltback also failed to work as intended: full analysis here

Conclusions

The video doesn't show any hardware fault or something that would be specific to that V13 unit.
Aside from the firmware bug related to alerts, the wheel behaved as expected considering the speed and acceleration observed.

At the moment, the V13 is not safe to ride at high speed due to a buggy alarm system

 

Note that the situation is still developing and findings from the logs could reveal additional issues compounding with the lack of beeps.

Edited February 17, 2023 by supercurio

[mrelwood]

 

3 hours ago, supercurio said:

What happened?

1. The wheel alerted the rider when reaching 72.1 km/h during strong initial acceleration
2. The rider eased up on his forward lean and acceleration following the alert
3. Then resumed accelerating since the wheel stopped beeping anymore.
4. When reaching 81.2 km/h, the wheel overleans. No beep, no tiltback but a crash instead.

That’s not quite what the rider said:

 

 

He also said that he did not slow down manually before the crash.

Edited February 16, 2023 by mrelwood

[Jean eRide.ie Community]

41 minutes ago, mrelwood said:

 

That’s not quite what the rider said:

 

 

He also said that he did not slow down manually before the crash.

I think he didn't stop leaning alright, the slower acceleration was caused by slow tilt back up to - 5 degrees angle, which he also didn't notice, but Inmotion confirmed. I'm sure he's updating the post to be more accurate. 

Edited February 16, 2023 by Jean eRide.ie Community

[supercurio]

@mrelwood it is hard to have a perfect recollection of events after a violent crash, with the shock and adrenaline rush.
We can see that the rider slowed down after the beeps, matching with the slow deployment of tiltback.

It is likely that if beeps would have been continuous instead of two really brief ones at 71.2 km/h, the rider would have had some chance to hear them.

Edited February 16, 2023 by supercurio

[Jean eRide.ie Community]

1 minute ago, supercurio said:

@mrelwood it is hard to have a perfect recollection of events after a violent crash, with the shock and adrenaline rush.
We can see that the rider slowed down after the beeps, matching with the slow deployment of tiltback.

It is likely that if beeps would have been continuous instead of two really brief ones at 71.2 km/h, the rider would have had a chance to hear them.

Sure, but I doubt he would have heard them at that speed to be honest, unless it is louder than on the EX20S, which is the loudest I've heard. 

[supercurio]

@Jean eRide.ie Community He also mentioned that he got no beep in his helmet, although expecting some. I asked the rider but couldn't get more details from his app alarms setup.

This is something that needs to become a lot easier and reliable, with safe and just working defaults on all apps.

[mrelwood]

5 minutes ago, supercurio said:

@Jean eRide.ie Community He also mentioned that he got no beep in his helmet, although expecting some. I asked the rider but couldn't get more details from his app alarms setup.

I thought he meant that he didn’t hear the beeps inside the helmet (where his ears were), not that he had a headphone system installed. 

5 minutes ago, supercurio said:

This is something that needs to become a lot easier and reliable, with safe and just working defaults on all apps.

Absolutely. 3rd party apps and two separate Bluetooth connections should not be expected.

[NSFW]

Speed-sensitive beep volume seems like a good idea too. All I hear in that crash video is wind noise.

My S22 is obnoxiously loud at a standstill, but barely audible in motion. There is no single volume level that works. But the software knows how fast the wheel is turning, so it should be pretty trivial to make the loudness proportional to the speed. 

[Punxatawneyjoe]

 

[supercurio]

Dawn Champion had a live stream to talk about the V13 cutout. I had the opportunity to ask about the V13 tiltback in her experience.
Like many others, she is a rider with interest in speed and the V13 is the first Inmotion wheel she rode.

It shows how for a rider who doesn't have a long experience with Inmotion wheels, the current tiltback is effective only if the power pads setup phsically prevents to lean further.

Here are the questions and answers:

Question 1:

Quote

How would you describe the V13 tiltback. In the data it seems linear and gradual, can you tell that it's happening immediately or after a couple seconds when it's a huge angle.

 

Question 2:

Quote

Thanks for describing tiltback, it sounds like it's too unobtrusive to act as instant feedback compared to the KS-16X one that's "tactile" and you can tell instantly

 

Edited February 16, 2023 by supercurio

[supercurio]

I changed the topic from "results in" to "contributes to" to reflect better the multiple factors in play that resulted in this overlean crash.

[supercurio]

From what Mark Yu said and thanks to his insider information, there is a significant insight to take into account for riders, it's in this message:

Quote

The speed alarm beeps is much shorter. But others beeps just have 5 seconds interval. Bad design

What this means:

• Beeps triggered by reaching the max speed (dynamic value) beeps will repeat quickly and seem continuous
• Beeps triggered by hitting another threshold like max torque, max phase current, max battery current, max inverter load (PWM %) will block others for a duration of 5 seconds.
• It is unclear if the 5s delay will block the max speed beeps as well, or only the same type of alarm.

 

Edited February 16, 2023 by supercurio

[alcatraz]

With a 140km/h pickup speed you'd think 80km/h should have been well within its limits. I wonder if the wheel just turned off to protect itself in which case you need an even greater safety margin than other brands to avoid doing that. Is that a fair assessment?

[Uras]

Just now, alcatraz said:

With a 140km/h pickup speed you'd think 80km/h should have been well within its limits

could it be, possibly, that it is much easier for the system to spin that wheel at 140km/h while it's in the air with no load at all - not even it's own weight? I imagine that it would struggle to do 80kmh up a steep hill too.

[alcatraz]

20 minutes ago, Uras said:

could it be, possibly, that it is much easier for the system to spin that wheel at 140km/h while it's in the air with no load at all - not even it's own weight? I imagine that it would struggle to do 80kmh up a steep hill too.

Yes of course. There's basically zero load, flat out max speed. Traditionally that has been a measure to extrapolate what a real world safe speed would be. However, if the wheel has additional safety margin (for itself, not rider) then you'd need to take an even larger chunk off the free spin speed to stay below.

Then there's the issue of HS or HT motors. If the V13 is an HS model then perhaps there isn't much torque at low speeds to begin with. Whatever max torque you can get at really low speeds, you'll have around half of that at half of the max speed (70km/h). 

I think they set the wheel self protection too conservatively. Some might argue it should be disabled because it puts the rider at risk. 

Either that or the HS motor doesn't have much max torque at low speeds. 

Edited February 17, 2023 by alcatraz

[mrelwood]

5 hours ago, alcatraz said:

With a 140km/h pickup speed you'd think 80km/h should have been well within its limits.

And speedwise it was. 

5 hours ago, alcatraz said:

I wonder if the wheel just turned off to protect itself

It didn’t, the wheel remained on during the crash. I guess you didn’t watch the video?

5 hours ago, alcatraz said:

Traditionally that has been a measure to extrapolate what a real world safe speed would be.

A “safe speed” doesn’t mean that you can lean into the pads as hard as you can and accelerate “as fast as possible” like the rider said that he tried. The wheel can only provide a limited amount of power to the motor, just like all wheels, now and in the future.

5 hours ago, alcatraz said:

If the V13 is an HS model then perhaps there isn't much torque at low speeds to begin with.

There’s a huge amount of torque available. But even a huge amount ends at some point.

5 hours ago, alcatraz said:

I think they set the wheel self protection too conservatively.

There are no signs for that to be honest. Several people have accelerated the V13 very fast at all kinds of speeds without issues. It’s one the most powerful wheels ever built.

5 hours ago, alcatraz said:

Some might argue it should be disabled because it puts the rider at risk.

To be able to overlean a V13 has had to have been a deliberate “all in” attitude, which makes the rider himself the risk. If anyone else as well thinks that the wheels are not capable of being overleaned, good luck to them. But I hope this event will open their eyes.

5 hours ago, alcatraz said:

Either that or the HS motor doesn't have much max torque at low speeds. 

It does, it has lots of torque. But 80km/h is definitely not a “low speed”, it’s near the max speed!

[supercurio]

Inmotion shared their official response with the rider who crashed.

I find the analysis spot on based on the video and data log. The conclusion confirms what both Mark Yu from Inmotion and I independently understood regarding the 5 seconds delay between alerts.

Likewise, the resolution steps proposed are the best that can be done with the current hardware. Really happy with how it has been handled.
It also means that nearly all Inmotion wheels, going back to at least V10F but possibly V5 as well will need a firmware update to remove the 5s alarm delay on all triggers. Watch out for these, and don't hesitate reminding Inmotion if it doesn't become available for your wheel.

 

Here's the data log for reference, highlighting the current alarm condition in red/yellow and final tiltback angle in red/green

 

Edited February 17, 2023 by supercurio

[Rawnei]

14 minutes ago, supercurio said:

Inmotion shared their official response with the rider who crashed.

I find the analysis spot on based on the video and data log. The conclusion confirms what both Mark Yu from Inmotion and I independently understood regarding the 5 seconds delay between alerts.

Likewise, the resolution steps proposed are the best that can be done with the current hardware. Really happy with how it has been handled.
It also means that nearly all Inmotion wheels, going back to at least V10F but possibly V5 as well will need a firmware update to remove the 5s alarm delay on all triggers. Watch out for these, and don't hesitate reminding Inmotion if it doesn't become available for your wheel.

 

Here's the data log for reference, highlighting the current alarm condition in red/yellow and final tiltback angle in red/green

 

Great response and understandment of the problem.

Would like to pinch all the victim blamers on their ears, stop it.

[alcatraz]

31 minutes ago, mrelwood said:

It didn’t, the wheel remained on during the crash. I guess you didn’t watch the video?

So is it safe to assume that it got pushed to its actual limit?  How can you see that motor engagement stayed on? Did the wheel spin up to 140 in the logs while tumbling?

32 minutes ago, mrelwood said:

There’s a huge amount of torque available. But even a huge amount ends at some point.

A begode motor free spinning out at 100 can be ridden carefully (without headwind) at up to 80. 140km/h free spin for the V13 and fail at 80 (adimittedly under more lean) leaves a big gap. Accounting for the lean still doesn't explain the big gap. Maybe it didn't have a lot of charge or something. Even then... it's only at 60% of it's free spin speed. It's like an RS overleaning at 58km/h. Master at 68km/h.

[Punxatawneyjoe]

44 minutes ago, supercurio said:

I find the analysis spot on based on the video and data log.

So basically they said, sorry we made it idiot proof but we forgot to make it stupid idiot proof, looks like now we have to do that .

Edited February 17, 2023 by Punxatawneyjoe
fogot it

[alcatraz]

55 minutes ago, mrelwood said:

It does, it has lots of torque. But 80km/h is definitely not a “low speed”, it’s near the max speed!

I think you misunderstood. I'll repeat. If torque for a motor is max near 0km/h and zero at 140, and we assume linear torque from the two points then at around half the speed you have around half the torque. 

[Rawnei]

6 minutes ago, Punxatawneyjoe said:

So basically they said, sorry we made it idiot proof but we forgot to make stupid idiot proof, looks like now we have to do that .

No what they said is that they made it safe but they forgot to actually make it safe and now they need to fix that.

[Punxatawneyjoe]

Just now, Rawnei said:

No what they said is that they made it safe but they forgot to actually make it safe and now they need to fix that.

It's strange that you come to that conclusion when the fact is that the rider admittedly had a 3rd party app running connected to the wheel and did not hear the FIRST set of beeps and then ignored the physical tilt back and continued to force the wheel forward causing the cut out. and it's inmotions fault because of a delay in the beeps that the rider never heard in the first place.

[Rawnei]

Just now, Punxatawneyjoe said:

It's strange that you come to that conclusion when the fact is that the rider admittedly had a 3rd party app running connected to the wheel and did not hear the FIRST set of beeps and then ignored the physical tilt back and continued to force the wheel forward causing the cut out. and it's inmotions fault because of a delay in the beeps that the rider never heard in the first place.

You have the inmotion analysis with fixes before your very eyes, what more do you need?

[supercurio]

12 minutes ago, Punxatawneyjoe said:

So basically they said, sorry we made it idiot proof but we forgot to make it stupid idiot proof, looks like now we have to do that .

I think that yes they tried to make it idiot proof and it works for most riders with a fairly restrictive power pads configuration, including the stock ones.
It prevents riders to lean further when the wheel tilts back. It pushes back on their legs and their speed feels capped since they can't move their center of gravity ahead more. You know in terms of biomechanics 

By doing that, making the tiltback gradual, they made it great because it's not jarring and sketchy - but they made the tiltback also imperceptible to many as a safety alert signal, and that becomes a problem in case the rider setup allows them to keep leaning just fine despite the change in angle - which was so gradual and slow that they couldn't tell it was happening. It could have been a change in wind speed for instance: impossible to tell.

I've been chatting with a multitude of V13 riders today and all of them were unable to identify as such, here are quotes:

Quote

I didn’t feel any sort of tiltback at 50mph. I started to feel it from 53mph. But i was already struggling before the tiltback came.

Quote

I was able to hit 80kmh but it is very hard to accelerate after 70kmh, the wheel starts to resist against you, definitely not a pleasant experience
there isnt as much torque as a master pro, 0-60kmh feels fast, after 70kmh the wheel stops assisting you, and tilts the pedals back

- the second rider understood during the discussion

So.. interesting! A tiltback that's so good that it becomes a problem.