Accelerometer vs Gyro for Stabilization

Hitec robotics including ROBONOVA humanoid, HSR-8498HB servos, MR C-3024 Controllers and RoboBasic
7 postsPage 1 of 1
7 postsPage 1 of 1

Accelerometer vs Gyro for Stabilization

Post by cbhanukiran » Fri Dec 10, 2010 11:04 am

Post by cbhanukiran
Fri Dec 10, 2010 11:04 am

Hi everyone,

I'm working on "Autonomous Mobile Robot Navigation" using Robonova-I. I want RN to avoid obstacles and navigate autonomously. So I would like to detect if RN falls due to slippery or something while walking, so that I can execute "stand_up" routine. Now what is a better choice ???

Use Gyro to avoid jerky motions or use Accelerometer to detect the abrupt motion ?

The solution that comes to my mind is "Gyro", since usage of Accelerometer involves some calculation and hence would be a bit slow, where as Gyro does it at the Hardware level.

Suggestions are very much appreciated.

Regards,
C BhanuKiran



,
Hi everyone,

I'm working on "Autonomous Mobile Robot Navigation" using Robonova-I. I want RN to avoid obstacles and navigate autonomously. So I would like to detect if RN falls due to slippery or something while walking, so that I can execute "stand_up" routine. Now what is a better choice ???

Use Gyro to avoid jerky motions or use Accelerometer to detect the abrupt motion ?

The solution that comes to my mind is "Gyro", since usage of Accelerometer involves some calculation and hence would be a bit slow, where as Gyro does it at the Hardware level.

Suggestions are very much appreciated.

Regards,
C BhanuKiran



,
cbhanukiran offline
Robot Builder
Robot Builder
Posts: 23
Joined: Thu Feb 25, 2010 7:20 am

Post by i-Bot » Fri Dec 10, 2010 11:26 am

Post by i-Bot
Fri Dec 10, 2010 11:26 am

A gyro reponds to angular velocity and on the Robonova the standard gyro or gyros are integrated direct to the Robonova software to provide stabilisation against small changes in body angle.
An accelerometer measures acceleration, either of the robot itself or due to gravity. In the robonova the accelerometer is used from robobasic to determine the orientation of the robot with respect to gravity.

If you want to stabilise to minimise falling you need a gyro, but it will never be very good. The gyro cannot detect orientation, so is no good to detect a fallen robot.

To detect a fallen robot and do the stand up, you need accelerometers. The calculation is fairly trivial, just a few compares.
A gyro reponds to angular velocity and on the Robonova the standard gyro or gyros are integrated direct to the Robonova software to provide stabilisation against small changes in body angle.
An accelerometer measures acceleration, either of the robot itself or due to gravity. In the robonova the accelerometer is used from robobasic to determine the orientation of the robot with respect to gravity.

If you want to stabilise to minimise falling you need a gyro, but it will never be very good. The gyro cannot detect orientation, so is no good to detect a fallen robot.

To detect a fallen robot and do the stand up, you need accelerometers. The calculation is fairly trivial, just a few compares.
i-Bot offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1142
Joined: Wed May 17, 2006 1:00 am

Re : Accelerometer vs Gyrometer for Stabilization

Post by cbhanukiran » Fri Dec 10, 2010 2:27 pm

Post by cbhanukiran
Fri Dec 10, 2010 2:27 pm

Yeah. Thanks a ton i-Bot.

Now I have yet another query. I have added another servo motor (in robot's chest cavity) and fixed a Camera to it (instead of head). So this (together) added quite some weight to the upper half.

I tried a lot to make RN (by catch and play) "stand_up", if it falls front/down, but failed. Do you have any "roboBASIC" routines to make the RN get up (with extra weight) ?

Is there any other way to determine RN moves using some Inverse Kinematics or some other stuff ?

That would be of great help.

C BhanuKiran
Yeah. Thanks a ton i-Bot.

Now I have yet another query. I have added another servo motor (in robot's chest cavity) and fixed a Camera to it (instead of head). So this (together) added quite some weight to the upper half.

I tried a lot to make RN (by catch and play) "stand_up", if it falls front/down, but failed. Do you have any "roboBASIC" routines to make the RN get up (with extra weight) ?

Is there any other way to determine RN moves using some Inverse Kinematics or some other stuff ?

That would be of great help.

C BhanuKiran
cbhanukiran offline
Robot Builder
Robot Builder
Posts: 23
Joined: Thu Feb 25, 2010 7:20 am

Post by i-Bot » Fri Dec 10, 2010 6:57 pm

Post by i-Bot
Fri Dec 10, 2010 6:57 pm

I have a head servo and a fairly heavy head with LCD display, but still use the standard get up routines.
When I have had problems on other robots to stand up, it has either been due to friction (especially on carpet) when sliding hands rather than a straight lift, or not having enough arm length to tip the centre of mass of the robot over the pivot point.
I suggest you remove the camera and observe very carefully the standard stand up to understand how the centre of mass shifts during the move, this should help you to modify that move.
The move is actually quick complex in the weight shift and not easy to achieve through catch and play because it depends on the dynamic movement. On mine it does not work if I step throught the individual moves, but works as a complete motion.

The only alternative environment I have seen was this:
http://web.mac.com/micono/RZE/RZ1Action.html
but never tried it.
I have a head servo and a fairly heavy head with LCD display, but still use the standard get up routines.
When I have had problems on other robots to stand up, it has either been due to friction (especially on carpet) when sliding hands rather than a straight lift, or not having enough arm length to tip the centre of mass of the robot over the pivot point.
I suggest you remove the camera and observe very carefully the standard stand up to understand how the centre of mass shifts during the move, this should help you to modify that move.
The move is actually quick complex in the weight shift and not easy to achieve through catch and play because it depends on the dynamic movement. On mine it does not work if I step throught the individual moves, but works as a complete motion.

The only alternative environment I have seen was this:
http://web.mac.com/micono/RZE/RZ1Action.html
but never tried it.
i-Bot offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 1142
Joined: Wed May 17, 2006 1:00 am

Post by customvinyl22 » Mon Jan 17, 2011 7:18 am

Post by customvinyl22
Mon Jan 17, 2011 7:18 am

thanks for sharing
thanks for sharing
customvinyl22 offline
Newbie
Newbie
Posts: 1
Joined: Mon Jan 17, 2011 7:17 am

Post by Albion Adam » Wed Jan 18, 2012 9:53 am

Post by Albion Adam
Wed Jan 18, 2012 9:53 am

I agree with you,, i also like it in the first time!
I agree with you,, i also like it in the first time!
Albion Adam offline
Newbie
Newbie
Posts: 1
Joined: Wed Jan 18, 2012 9:52 am

Post by ozfiddler » Sat Jan 21, 2012 12:10 am

Post by ozfiddler
Sat Jan 21, 2012 12:10 am

Hi C BhanuKiran,

I'd be very interested to know how you are handling the images from the camera/head you have installed - can you "see" from a remote computer? I added a servo in the chest cavity with a distance sensor for a head and was trying to navigate around an obstacle course, but my Robo kept getting lost! I've got a compass module but haven't got around to working out a way to use it without interference from the servos.

Anyway, sounds like a great project and I'd be keen to hear more about your progress.

Cheers,

Alan
Hi C BhanuKiran,

I'd be very interested to know how you are handling the images from the camera/head you have installed - can you "see" from a remote computer? I added a servo in the chest cavity with a distance sensor for a head and was trying to navigate around an obstacle course, but my Robo kept getting lost! I've got a compass module but haven't got around to working out a way to use it without interference from the servos.

Anyway, sounds like a great project and I'd be keen to hear more about your progress.

Cheers,

Alan
ozfiddler offline
Savvy Roboteer
Savvy Roboteer
Posts: 93
Joined: Tue Oct 26, 2010 1:01 am


7 postsPage 1 of 1
7 postsPage 1 of 1
cron