Accelerometer

[netbug]

Hi all,

I am starting a project that uses a 3 axis accelerometer, to measure the accelaration of a RC car. The problem is when the car is stopped it measures gravity, and I can't align one axis with gravity so I have gravity mainly on one axis. Every time I make a small move the gravity changes and reflects to all 3 axis.

The big question is how to measure accelaration when the car is running if gravity seems to spread randomly by the 3 axis ?

Best regards,

Pedro

 

[stevez]

My understanding of a 3 axis accelerometer is that it will provide information on the acceleration for the x, y and z axis. Each axis is defined by the sensor unless the sensor contains some additional intelligence to compensate for the orientation of the base sensors. If it is mounted to the chassis of an RC car and is sufficiently sensitive it will provide an output that reflects every minute change in velocity in all three directions. The part that might be missed is rotation - if mounted to an RC car that could occur on cornering or acceleration/deceleration if it has a suspension. If your point of reference is horizontal and vertical in terms of the earth then you'll need to manage that. One way to simplify things would be to perform measurements on a flat surface in one direction (in-line w/chassis of car), possibly a second (perpendicular to chassis) and make the assumption that the chassis remains level.

My recollection of RC cars is based on my son's experience - I'd see the cars bouncing all over the place on dirt tracks.

 

[MrMikey83]

I am having the same problem with my ballance bot. The accelorometer senses every little gravity shift including vibration. So when the motors are running, all I get from the sensor is garbage. It will be the same for you unless you can figure out how to filter out the unwanted noise.
You need to be able to filter out the short bursts (vibrations) while still keeping the acceloration signal steady.
~Mike

 

[spindrah]

I am having the same problem with my ballance bot. The accelorometer senses every little gravity shift including vibration. So when the motors are running, all I get from the sensor is garbage. It will be the same for you unless you can figure out how to filter out the unwanted noise.
You need to be able to filter out the short bursts (vibrations) while still keeping the acceloration signal steady.
~Mike

Hey MrMikey

you still don't believe me about the accelorometer not being able to do this?
try and graph the output of the accel as your bot freefalls from vertical to horizontal.
you will see that it reads near 0 untill it hits the floor, then it will spike!

or another thing you can try.
hang it by a string and let it act as a pendulum, graph will be a flat line(not the sinwave you may think)

 

[ljcox]

I would have thought it would read g while in free fall and o while stationary. Are you sure it measures accel? Perhaps it measures force.

Len

 

[MrMikey83]

Well, the vibrations make the accel spike just as if it hit the floor. On my bot, the vibrations from the motor cause the accel to sense the small back and forth jerking motions of the vibration. When I put this input to the motors, the jerking from the vibrations transmits to the motors making jerky movements and makes things even worse lol

The problem I am having with my bot is that the accel is way too sensative and picks up the tiniest of vibrations. I need to find a way to filter out the tiny movements and only focus on gravity. On my bot, the accel should only sense gravity downward to correct for drift on the gyro which is the sensor for tilting rate. I hope I made sanse...it makes sense in my head but hard to transfer to text. heh
~Mike

 

[Pommie]

MrMickey,
How will your bot know the difference between standing still and traveling forward at a constant velocity. Your accelerometer will read the same in both instances.

Spindrah,
Surely, the radial component will be sinusoidal due to centrifugal force and the tangential component will be a sinusoidal due to the sideways movement. Are you saying that the resultant of the two will be constant and equal to gravity?

Netbug,
With all accelerometers, if the object is moving at a constant velocity in a straight line then the only acceleration will be gravity. Your 3 axis accelerometer is acting as a spirit level. If you resolve the 3 readings into 1 vector then you should find it is G long and in a downward direction.


Mike.

 

[ljcox]

How will your bot know the difference between standing still and traveling forward at a constant velocity.
Mike.

Travelling needs 2 Ls. Consider hope & hoping, hop and hopping.

Len

 

[MrMikey83]

Well, if the motors are running forward, the bot already knows it is moving forwrd. The only thing I need to sense is rate of tilt so that if it start falling over, it can put the wheels back under the bot by speeding them up.
When the bot tilts, the gyro provides a rate, but when the tilt stops, the gyro goes back to zero, even if the bot is tilted at 45 degrees, the gyro will read 0 rate of tilt is its not moving. This is where the accel comes in. While it is tilted at 45 degrees, the accel can tell the bot that gravity is in the wrong place and it need to correct itself.
~Mike

 

[Nigel Goodwin]

Nice theory Mike?.

But you've been working on this for a LONG! time now, and it doesn't sound like you're getting very far?.

 

[MrMikey83]

With school and work, I haven't had as much time as I'd like to be able to work on this project. Every time I go to work on school work on my computer, this bot is right next to the monitor staring me in the face. lol

The biggest problem I'm having is filtering out the vibrations. If I can get over that hurdle, I think I will make much more progress (if time allows).
I have several teachers who I have shown the bot to and they all want to see it working...maybe if I ask them for less homework in exchange for letting them see it finished hehe.
~Mike

 

[spindrah]

When the bot tilts, the gyro provides a rate, but when the tilt stops, the gyro goes back to zero, even if the bot is tilted at 45 degrees, the gyro will read 0 rate of tilt is its not moving

yes the gyro goes bake to 0, but you should have been integrating the gyro.
then you would know its at 45 degrees

say the gyro reads + when falling forward, and - when falling back
if you get 5 readings say 10,15,8,3,0
you have +36 in your variable, and you know not to stop correcting just because the last reading was 0
say the next reading is -13, you still have +23 in your variable,so you're still leaning forward

as for the accel integrat it and use it to adjust the gyro's drift
the average of 10-30 readings will have very little noise in it

 

[JohnBrown]

How will your bot know the difference between standing still and traveling forward at a constant velocity.
Mike.

Travelling needs 2 Ls. Consider hope & hoping, hop and hopping.

Len

Not if you're American. They only believe in one L. And one 'eaven, probably.

 

[Roff]

Never mind. John Brown beat me to it, and he was also funny.
BTW, do Brits believe there is more than one 'eaven?

 

[Nigel Goodwin]

Never mind. John Brown beat me to it, and he was also funny.
BTW, do Brits believe there is more than one 'eaven?

I suspect most probably don't believe in one at all?.

 

[john1]

Hi,

Yes most sensible Englishmen believe in more than one heaven.
However Catholics have to be segregated in their heaven,
cos they think thats the only one there is.

'Witnesses' here come round knocking on doors, trying to get
converts by verbally trying to convince people that they should
convert to being 'Witnesses'
They claim that this is heaven, and that soon it will get 'all
tidied up' by supernatural intervention,
And only populated by 'Witnesses'.

I am of course not sure which way to go, but i feel there are
many heavens that comprise the whole, and that one ends up in
the place that one deserves, which may or may not be the place
one wants. (just an opinion)

John