Legacy Forum: Preserving Nearly 20 Years of Community History - A Time Capsule of Discussions, Memories, and Shared Experiences.

Premium stability test

Bioloid robot kit from Korean company Robotis; CM5 controller block, AX12 servos..
33 postsPage 3 of 31, 2, 3
33 postsPage 3 of 31, 2, 3

Post by cajacava » Sat May 22, 2010 4:35 am

Post by cajacava
Sat May 22, 2010 4:35 am

Thanks for your help FavinovX, I try to let the robot in the ground ans then run the program, in this point all was fine, when I move the robot to change it position then happens what I wrote before, It sends the two legs and foot to the same side and I need to turn off quicky befofe something bad happens. How can I calibrate my robot?, is correctly builded and if I run the demo program its fine.

Thanks

Carlos

FabinovX wrote:Maybe, that's due to the noise of AX12 position feedback.
you can try to enlarge dead zone in the soft.
but first you may try this :
Just put the robot on the ground before starting the program.
if it still goes crazy, then you should calibrate your robot.
In effect, if your ax12 can't reach their goal position when standing there will be a difference between goal and present position, so that, the robot will analyse this and move for nothing.

EDIT : the gyro is just use to stabilize when there is quick movements,
it doesn't help when you do slow movements.
it's only AX12 position feedback that helps in this case.
for information, my very first try was based on AX12 torque feedback but it was more noisy than position feedback and the soft was really more complicated and longer because torque value changes according to the rotation direction too.

Hope it will help.

Cheers
FabinovX
Thanks for your help FavinovX, I try to let the robot in the ground ans then run the program, in this point all was fine, when I move the robot to change it position then happens what I wrote before, It sends the two legs and foot to the same side and I need to turn off quicky befofe something bad happens. How can I calibrate my robot?, is correctly builded and if I run the demo program its fine.

Thanks

Carlos

FabinovX wrote:Maybe, that's due to the noise of AX12 position feedback.
you can try to enlarge dead zone in the soft.
but first you may try this :
Just put the robot on the ground before starting the program.
if it still goes crazy, then you should calibrate your robot.
In effect, if your ax12 can't reach their goal position when standing there will be a difference between goal and present position, so that, the robot will analyse this and move for nothing.

EDIT : the gyro is just use to stabilize when there is quick movements,
it doesn't help when you do slow movements.
it's only AX12 position feedback that helps in this case.
for information, my very first try was based on AX12 torque feedback but it was more noisy than position feedback and the soft was really more complicated and longer because torque value changes according to the rotation direction too.

Hope it will help.

Cheers
FabinovX
cajacava
Newbie
Newbie
Posts: 2
Joined: Wed May 05, 2010 3:52 am

Post by FabinovX » Mon May 24, 2010 11:56 am

Post by FabinovX
Mon May 24, 2010 11:56 am

A Robotis calibration tutorial link can be found here :
http://robosavvy.com/forum/viewtopic.php?t=5977

but if your robot doesn't go crazy when you let it on the ground, it won't help you i think.

the problem with AX12 feedback is that you have sometimes strange peak values that comes.
if you have your robot in your hand, this peak value (when it comes) stay for too much time and the program starts to move the legs according this fake value.
that's why the robot can go crazy.

if your robot is let on a plane surface, when this fake peak value comes, the program start to move the legs too, but this feedback value will quickly be replaced by a unnoisy value, and so, the robot won't get crazy.

keep in mind this program was made for fun, it was one of my first.
the more the actuators feedback is perfect, the more program works fine,
but if it is too noisy, you will face such inconvinience as you encountered.

try to enlarge deadzone values in the program.
your robot will be less reactive but it may help to resolve your problem.

in all case, the best way to do such a program really run fine is too use a 2 axis accelerometer in addition with the gyro.
i think i will buy some but i would prefer going with an AX-S20 like sensor.
but for the moment, there is no alternative choice to the robotis AX-S20.
the 6 axis imu from hylands seems not to work fine with cm510 according to some people on this forum.....

EDIT: in order to protect your hardware, you can define limit values for each actuators using roboplus manager.
if you do so, the actuator will go to alarm mode before you break something.
The more i use AX12, the more i think the only reliable feedback value is the temperature :)
A Robotis calibration tutorial link can be found here :
http://robosavvy.com/forum/viewtopic.php?t=5977

but if your robot doesn't go crazy when you let it on the ground, it won't help you i think.

the problem with AX12 feedback is that you have sometimes strange peak values that comes.
if you have your robot in your hand, this peak value (when it comes) stay for too much time and the program starts to move the legs according this fake value.
that's why the robot can go crazy.

if your robot is let on a plane surface, when this fake peak value comes, the program start to move the legs too, but this feedback value will quickly be replaced by a unnoisy value, and so, the robot won't get crazy.

keep in mind this program was made for fun, it was one of my first.
the more the actuators feedback is perfect, the more program works fine,
but if it is too noisy, you will face such inconvinience as you encountered.

try to enlarge deadzone values in the program.
your robot will be less reactive but it may help to resolve your problem.

in all case, the best way to do such a program really run fine is too use a 2 axis accelerometer in addition with the gyro.
i think i will buy some but i would prefer going with an AX-S20 like sensor.
but for the moment, there is no alternative choice to the robotis AX-S20.
the 6 axis imu from hylands seems not to work fine with cm510 according to some people on this forum.....

EDIT: in order to protect your hardware, you can define limit values for each actuators using roboplus manager.
if you do so, the actuator will go to alarm mode before you break something.
The more i use AX12, the more i think the only reliable feedback value is the temperature :)
FabinovX
Savvy Roboteer
Savvy Roboteer
Posts: 74
Joined: Mon Feb 01, 2010 4:12 pm

Post by SK » Mon May 24, 2010 9:01 pm

Post by SK
Mon May 24, 2010 9:01 pm

The strange peak values you describe sound very much like worn out potentiometers in the AX-12 servos. I know these symptoms from heavily used RX-28s, they don't appear with new servos though.
The strange peak values you describe sound very much like worn out potentiometers in the AX-12 servos. I know these symptoms from heavily used RX-28s, they don't appear with new servos though.
SK
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 105
Joined: Mon Jun 26, 2006 1:00 am
Location: near Frankfurt/Main, Germany

Previous
33 postsPage 3 of 31, 2, 3
33 postsPage 3 of 31, 2, 3