Gyro for Self Leveling?

This site may earn a commission from merchant affiliate
links, including eBay, Amazon, and others.
Somehow missed the last 6 responses before i posted this... but here it is anyway:

-------------------------------
i am a quad guy and i also don't see this working...

the biggest problem is not the PID as i believe you could make something capable of handling break/acceleration to maintain the pitch of the car...

the biggest problem is that the car doesn't know when it is in the air and therefore when to apply these corrections... every time you drive up a hill the car will think it is pitched up and will apply the brakes to try to correct that pitch. when the car doesn't pitch forward (because it is going up a hill) it will add more and more brake as the PID winds up until it comes to a complete stop (might even go backwards violently). And forget trying to do wheelies or backflips...

This works on a racing quad because you command the angle with the sticks and the PID just maintains the last commanded angle. On a car, you just tell it to go forwards and backwards and the angle is incidental to other external factors (terrain/ramps etc...).

There are other problems, but the above are showstoppers...
 
Last edited:
I'm surprised this thread was dug from the archives.. 🤣... It was never about multi axis stabilization.. A heli gyro and a multi rotor gyro are two different animals.. A heli gyro works on one axis and controls one channel, the rudder...

No reprogramming of esc needed, it would control the throttle/ brake in the same fashion as a rudder if the gyro was mounted horizontal instead of vertical...

I'm not saying this is a functional idea, but would it control throttle and brake with yaw on that axis? Yes, very much so..
It worked with a the yaw of an r/c submarine, 25 years ago.
 
It worked with a the yaw of an r/c submarine, 25 years ago.
Yes because you have a dedicated yaw control surface and it can be active all the time. The big obstacles here are that the RC needs a way to know when to activate (only when in the air on a “big” jump amd it also has a control method that is used for throttle/brake and not independent.
 
Detecting a jump may not be that hard as you're effectively in freefall, which is quite easy to spot with an accelerometer so you could trigger once you've been in freefall for more than a certain time. You detect freefall by noticing that there's apparently no gravity acting on the accelerometer, which gives you a different problem: you've got no way to detect which way "up" is, so in order to level out, you'd need to know your pitch at the time of launch, and then integrate rotation whilst in flight. Knowing your pitch at the time of launch isn't easy either, because the car is constantly accelerating when on the ground.
 
Detecting a jump may not be that hard as you're effectively in freefall, which is quite easy to spot with an accelerometer so you could trigger once you've been in freefall for more than a certain time. You detect freefall by noticing that there's apparently no gravity acting on the accelerometer, which gives you a different problem: you've got no way to detect which way "up" is, so in order to level out, you'd need to know your pitch at the time of launch, and then integrate rotation whilst in flight. Knowing your pitch at the time of launch isn't easy either, because the car is constantly accelerating when on the ground.


Accelerometer... that's a good idea... but you would need to immediately release the throttle on launch or the added rotational acceleration would make this unreliable... and you would need to filter out the high freequency vibrational noise of the wheels rotating, however it could probably resolve the intentional wheelie issue.

For determining what "level" is, you could record the pitch on initialization as your "level" just make sure you power the car up on level ground and have the PID return the car to that pitch on 0Gs.
 
I've built a drone before. You don't need a super expensive dji to get that level of control. Most flight controllers for the custom diy builds run on betaflight which is Programmable to get the motors to do whatever you want in response to gyro inputs. But out of the box they're programmed to keep it level in stab mode and respond to your stick inputs. This could be done for under $200.

The level of effort would be very high. Since I came from the drone world I've been trying to dream up something that can cross over to this platform. So I was already thinking about this idea but tucked it away once I realized how difficult it would be
 
you would need to immediately release the throttle on launch or the added rotational acceleration
Possibly. You only need the net linear acceleration. If the accelerometer is close to the car's center of rotation then this should still be close to zero even in the presence of angular acceleration.
For determining what "level" is, you could record the pitch on initialization as your "level" just make sure you power the car up on level ground and have the PID return the car to that pitch on 0Gs.
You don't even need to be on level ground. As long as you're not moving, and the gyro is mounted level in the car, gravity will tell you which way up you are.

The problem is knowing your attitude once you start moving. My understanding is that a gyro only gives you angular acceleration. If you want angular velocity, you need to integrate, and if you want attitude you need to integrate again, both of which will accumulate errors as time goes on. I'd expect that approach to last almost no time at all given the kind of accelerations that an RC car sees.
 
Has anyone tried to use a gyro for wheelie control, or to get the car to self level when jumping? I know the challenge of controlling the car is most of the fun, but it would also be cool to occasionally just send it and watch it stay perfectly level.
Japan has it I've seen it in a video years ago. I'm currently looking for it again.
 
Sorry to revive this but this is a cool project and lots of good ideas. So I though I would add something that was not mentioned (unless I missed it).

Quadcopter that does autonomous navigation use a barometer (baro) that’s how the quad can keep its altitude (could also be combined with gps for more accuracy) but the point I want to make is that the car (equipped with electronics) via a baro can “sense” when the car is in the air (it knows it’s altitude when on the ground) with some threshold that could be done by user through some config (let’s say if you “tell” the car that if it sense it is at more than 3 feet above its initial altitude, 95% chance you are in the air. That can be more accurately deducted with the accelerometer.
Once the wheels leave any surface (ramp, ground) acceleration will not be increasing in forward direction anymore (no traction from the ground).

To conclude. For a surface véhicule to “know” if it’s in the air is not that “difficult”. The same to know if the car is going towards the ground (and it’s time to do something cause a crash is about to happen) is the same way (combo of baro + acc).
Gyro will tell you how the car is spinning on the 3 axis.

Then “all needs to be done” is do whatever user do to control landing once the car senses its going towards the ground.

And then to push it. An accrobacy mode could easily take control once the car goes towards the sky (to impress your buddies) to make the car do some random positions flips and whatnot. Then once it goes down, take safe control for landing.

I don’t have experience in big jumps and how to control it.

Nobody mentioned parachute yet?
When the car finish its aerial show, trigger parachute for self landing!

Sky is really the limit!
 
Sorry to revive this but this is a cool project and lots of good ideas. So I though I would add something that was not mentioned (unless I missed it).

Quadcopter that does autonomous navigation use a barometer (baro) that’s how the quad can keep its altitude (could also be combined with gps for more accuracy) but the point I want to make is that the car (equipped with electronics) via a baro can “sense” when the car is in the air (it knows it’s altitude when on the ground) with some threshold that could be done by user through some config (let’s say if you “tell” the car that if it sense it is at more than 3 feet above its initial altitude, 95% chance you are in the air. That can be more accurately deducted with the accelerometer.
Once the wheels leave any surface (ramp, ground) acceleration will not be increasing in forward direction anymore (no traction from the ground).

To conclude. For a surface véhicule to “know” if it’s in the air is not that “difficult”. The same to know if the car is going towards the ground (and it’s time to do something cause a crash is about to happen) is the same way (combo of baro + acc).
Gyro will tell you how the car is spinning on the 3 axis.

Then “all needs to be done” is do whatever user do to control landing once the car senses its going towards the ground.

And then to push it. An accrobacy mode could easily take control once the car goes towards the sky (to impress your buddies) to make the car do some random positions flips and whatnot. Then once it goes down, take safe control for landing.

I don’t have experience in big jumps and how to control it.

Nobody mentioned parachute yet?
When the car finish its aerial show, trigger parachute for self landing!

Sky is really the limit!

You're Assuming a LOT by stating a barometer, gyro and GPS is all that's needed. These are JUST the sensors.. You also need Computing Power AND Custom software to analysis, process and control the motor and steering...

The timing and speed(quickness) to adjust a 15-20lb rig flying through the air traveling at 50+MPH and all it's got are rotating wheels to attempt to "gyro stabilize" itself through some random trajectories (in time) to properly land back onto its wheels -right side up...

That's just the basics..

Keep in mind that on some advanced drones -there are Five Pairs of stereoptic HD cameras to see objects that it coming towards it.

It'll need very complex algorithms to decipher the difference between clouds of dusts in the air and actual ground that it's heading to ,after it launches into the air.

I am sure I'm over simplifying all of these variables.. but you get the drift.



**************

Oh and Yes! On parachute 💪🪂
Here >>
https://www.arrmaforum.com/threads/functional-parachute-video.58471/
 
Last edited:
Yes I mean I don’t think the intent here is to avoid object while the car is in the air. The plan is not to build a bullet proof system for the army.

I think a first requirement is to land on its four wheels (or at least on its belly).
So all I am saying is for that, you only need a baro, acc and gyro for sensor and signals to control the servo and ESC.
Yes of course you need some cpu to process all that (a sensor by itself does not do much).

With the opensource flight controller already existing (the cpu power and sensors) it would be a good starting platform.

It’s a cool project I would be interested to participate.
 
I'm surprised this thread was dug from the archives.. 🤣... It was never about multi axis stabilization.. A heli gyro and a multi rotor gyro are two different animals.. A heli gyro works on one axis and controls one channel, the rudder...

No reprogramming of esc needed, it would control the throttle/ brake in the same fashion as a rudder if the gyro was mounted horizontal instead of vertical...

I'm not saying this is a functional idea, but would it control throttle and brake with yaw on that axis? Yes, very much so..

In the interest of making the discussion as accurate as possible, this is no longer true, since flybarless controllers came on the market for helis, maybe mid 2000's or so? Previously, with flybars, the gyro only controlled the rudder. With FBL, the gyro is 3-axis, and controls the rudder, as well as pitch and yaw.

When I was last active with them, there were a few controllers on the market that could also influence the throttle channel, as a proactive RPM governor. Rather than the ESC governor waiting to see the RPMs drop, before it could react and increase the throttle, these systems would instead also look at your command inputs. If you punch the collective, increasing the pitch of the rotors, to climb fast, it would realize that this is about to put a large load on the motor, so the controller could already start to call for more throttle, even as the servos just start to move. Making it more proactive.

I think this is an interesting discussion. It feels to me like this is a difficult challenge to solve, for many of the reasons already mentioned. Just touching on a few, as they occur to me (bearing in mind I've barely gotten to jump my truck yet, and haven't gotten to try correcting in the air):
- Gains would depend on gearing, motor performance, wheelbase, and tire weight, at a minimum. Having seen people say that tires X or Y allow very little control in the air, I have to think that some of these are pretty significant. So tuning sounds tricky.
- Its ability to work at all would depend on your throttle when you go in the air, I would think. If you're already at 100%, there's not much "more to give", from the controller, to make you nose-up.
- Figuring out when you're in the air doesn't seem trivial. Personally, it doesn't seem to me like GPS or a barometer would be enough. I like the idea of using the gyros, and accelerometers, to figure it out. There would be much more high frequency "noise" in the measurements when you're bouncing around over dirt. If you were to graph the data, I'll bet you'd see an awful lot of comparatively-small activity. When you suddenly go off a jump, the "large" acceleration would change a lot (gravity goes to basically 0), and since you're not bouncing around on grass/dirt, the small-scale stuff would also decrease a lot. Suddenly, you'd be mainly picking up stuff like tire vibrations. Maybe this change tells it "Airborne! Time to make corrections!".
- Even with all that stuff figured out, what do you want to do with all that? Keep the same angle, and land back-wheels-down, if that's the ramp angle you took off from? And whatever you decide to do, how quickly does this need to happen? The unit doesn't know how long you'll be in the air.

Even if this isn't any easy challenge, I still think it's an interesting discussion! And there are loads of clever people, I'll bet someone could come up with some type of solution.

And for one of the original questions, what about starting simpler? Something to let the vehicle pull, and hold, a 45 degree wheelie limit, or whatever? Sort of acting like a virtual wheelie-bar. This would sound more straightforward, at least, as you're still on the ground.
 
Last edited:
Yes I mean I don’t think the intent here is to avoid object while the car is in the air. The plan is not to build a bullet proof system for the army.

I think a first requirement is to land on its four wheels (or at least on its belly).
So all I am saying is for that, you only need a baro, acc and gyro for sensor and signals to control the servo and ESC.
Yes of course you need some cpu to process all that (a sensor by itself does not do much).

With the opensource flight controller already existing (the cpu power and sensors) it would be a good starting platform.

It’s a cool project I would be interested to participate.

Okay.
However, you're missing the core - a WORKING vehicle platform..
Someone needs to come up with a chassis that has four independent motors & ESCs > driving four independent wheels (of known weight).

So a possible "flight controller" can be allowed to control all four rotating mass (wheels) independently (similar to a quadcopter's four independent fan blades) to begin to make the necessary adjustments while 'inflight' (attempting Big air jumps.)
 
Okay but again we are not asking for 100% control in the air. There are scenario where control will be too costly (car goes too fast, or sideways in the air or just user fighting against the cpu commands … or even the sensor won’t keep up let alone the cpu).

I think basic “recovery from a jump” so that the car does land on its belly to minimise crash so that beginners can learn how to control their car after a jump.

It’s a bit like AVC. It sounds like experimented driver don’t like it because it remove some driver control but beginner appreciate it (like I do) because it remove some stress (my son don’t want to break the car every time we go out to play with it).

AVC has its limit and does not prevent all the accident. This “jump recovery” feature could be the same, it’s goal is not to prevent all the crash but make it more likely to land on its belly after a jump.

You don’t need to control all the wheels just the turn servo and the ESC. At a minimum shut down the speed before landing so the gears don’t loose teeth at landing for those (like my son) that only know one throttle position (full speed)!!!!
 
Old Thread: Hello . There have been no replies in this thread for 90 days.
Content in this thread may no longer be relevant.
Perhaps it would be better to start a new thread instead.
Back
Top