2 very ingenious Robobuilder Videos

Korean company maker of Robot kits and servos designed for of articulated robots. Re-incarnation of Megarobotics.
5 postsPage 1 of 1
5 postsPage 1 of 1

2 very ingenious Robobuilder Videos

Post by PedroR » Mon Feb 08, 2010 1:19 pm

Post by PedroR
Mon Feb 08, 2010 1:19 pm

Hi all

We have received teh news about two very ingenious videos involving robobuilder:

The first is a Robobuilder fitted with an ARM7 main controller and 5 degrees IMU with a Kalman filter:

phpBB [media]

http://www.youtube.com/watch?v=8adyyIIvb1Y


The second one is a Robobuilder that is capable of self balancing but has NO gyros and NO acceleration sensor, simply using feedback from his servos to keep himself stable.

phpBB [media]

http://www.youtube.com/watch?v=zXIRJMkWMCc

As a personal note, I must say I got this information from the manufacturer (Robobuilder) but I don't know how they confirmed the technical solution in the above video.

In any case, after some thought, I think the robot is evaluating the impact by analysing if there are deviations in the target position of the arm servos and leg servos.
I think that if you tweak teh PID settings to allow for more elasticity you can more easily detect these movements. I am not sure if this is what he did though. In any case it seems very ingenious.

Regards
Pedro
Hi all

We have received teh news about two very ingenious videos involving robobuilder:

The first is a Robobuilder fitted with an ARM7 main controller and 5 degrees IMU with a Kalman filter:

phpBB [media]

http://www.youtube.com/watch?v=8adyyIIvb1Y


The second one is a Robobuilder that is capable of self balancing but has NO gyros and NO acceleration sensor, simply using feedback from his servos to keep himself stable.

phpBB [media]

http://www.youtube.com/watch?v=zXIRJMkWMCc

As a personal note, I must say I got this information from the manufacturer (Robobuilder) but I don't know how they confirmed the technical solution in the above video.

In any case, after some thought, I think the robot is evaluating the impact by analysing if there are deviations in the target position of the arm servos and leg servos.
I think that if you tweak teh PID settings to allow for more elasticity you can more easily detect these movements. I am not sure if this is what he did though. In any case it seems very ingenious.

Regards
Pedro
PedroR offline
Savvy Roboteer
Savvy Roboteer
Posts: 1199
Joined: Mon Jun 16, 2008 11:07 pm

Post by Felix » Fri Feb 12, 2010 4:13 pm

Post by Felix
Fri Feb 12, 2010 4:13 pm

While it is still pretty clever stuff, in the second video it looks like most (if not all) of the impacts are not hard enough to knock the robot over and it is more or less stable again by the time it reacts. Just my two cents though!

Is the first one using an acceleration sensor or is it using the cameras?
While it is still pretty clever stuff, in the second video it looks like most (if not all) of the impacts are not hard enough to knock the robot over and it is more or less stable again by the time it reacts. Just my two cents though!

Is the first one using an acceleration sensor or is it using the cameras?
Felix offline
Robot Builder
Robot Builder
Posts: 12
Joined: Wed Jun 17, 2009 12:24 pm

Post by l3v3rz » Sat Feb 13, 2010 12:33 am

Post by l3v3rz
Sat Feb 13, 2010 12:33 am

Well - not really genius - but seemed at funny when I came across it.

"When you can't run, you crawl, and when you can't crawl, you find somebody to carry you."

phpBB [media]
Well - not really genius - but seemed at funny when I came across it.

"When you can't run, you crawl, and when you can't crawl, you find somebody to carry you."

phpBB [media]
l3v3rz offline
Savvy Roboteer
Savvy Roboteer
Posts: 473
Joined: Fri Jul 18, 2008 2:34 pm

Post by PedroR » Mon Feb 15, 2010 1:14 pm

Post by PedroR
Mon Feb 15, 2010 1:14 pm

l3v3rz: LOL. nice idea :)


Felix: the first one is using an IMU (which is a combination of Accelerometer and Gyro). I believe the first one is NOT using the RBC controller.

They probably replaced it with a more powerful controller (I think ARM but I am not sure) to able to run the Inverse Kinematics solution.
As I mentioned the sensors are a 6 axis IMU I think (3 axis gyro, 3 axis accelerometer).

The standard RBC controller box will not let you use an IMU unless you use on with a microprocessor, reprogramed to be able to sit on the bus an emulate itself as a device.


As I write this, I just thought of this recent product we have in our catalog: http://robosavvy.com/store/product_info ... cts_id/605

This is a 6 axis IMU with a re programmable ARM Cortex running at 64Mhz (it comes with a bootloader for easy programming).
The IMU is accessed via a UART TTL connection so you may actually be able to connect this one to a Robobuilder bus provided you modify the software to understand the protocol and act as a bus device.


We are waiting for a new revision of the Open Source software for this IMU that implements a Kalman Filter which is likely the best filtering you can have for stability processing.



For those who are really ambitious, you can actually use the ARM7 on the IMU to control the whole robot and just leave the RBC in place to offer battery, the bus breakout connectors and the Direct Control mode to have the PC talk to the bus and the IMU on the bus (but doing no LOGIC work).
I could go on and on about this solution which is a very cool one btw but I'll stop here.

It is actually a very clean and small solution for Robobuilder and I believe ti may fit in the chest of the robot.

Pedro.
l3v3rz: LOL. nice idea :)


Felix: the first one is using an IMU (which is a combination of Accelerometer and Gyro). I believe the first one is NOT using the RBC controller.

They probably replaced it with a more powerful controller (I think ARM but I am not sure) to able to run the Inverse Kinematics solution.
As I mentioned the sensors are a 6 axis IMU I think (3 axis gyro, 3 axis accelerometer).

The standard RBC controller box will not let you use an IMU unless you use on with a microprocessor, reprogramed to be able to sit on the bus an emulate itself as a device.


As I write this, I just thought of this recent product we have in our catalog: http://robosavvy.com/store/product_info ... cts_id/605

This is a 6 axis IMU with a re programmable ARM Cortex running at 64Mhz (it comes with a bootloader for easy programming).
The IMU is accessed via a UART TTL connection so you may actually be able to connect this one to a Robobuilder bus provided you modify the software to understand the protocol and act as a bus device.


We are waiting for a new revision of the Open Source software for this IMU that implements a Kalman Filter which is likely the best filtering you can have for stability processing.



For those who are really ambitious, you can actually use the ARM7 on the IMU to control the whole robot and just leave the RBC in place to offer battery, the bus breakout connectors and the Direct Control mode to have the PC talk to the bus and the IMU on the bus (but doing no LOGIC work).
I could go on and on about this solution which is a very cool one btw but I'll stop here.

It is actually a very clean and small solution for Robobuilder and I believe ti may fit in the chest of the robot.

Pedro.
PedroR offline
Savvy Roboteer
Savvy Roboteer
Posts: 1199
Joined: Mon Jun 16, 2008 11:07 pm

Post by badcommandorfilename » Wed Feb 24, 2010 8:38 am

Post by badcommandorfilename
Wed Feb 24, 2010 8:38 am

Hi everyone!

The second video in the OP is mine - it's actually an improved version of one of my earlier attempts at balance.

It works by obtaining feedback information from the ankle servos. If a force is applied in one of four directions, it will move with the force to avoid falling over. These videos probably show it better.

I've even made the source code available on my blog if you want to have a look.

I'm glad there are people out there who appreciate my hard work! I will try to make some more for you to enjoy.
Hi everyone!

The second video in the OP is mine - it's actually an improved version of one of my earlier attempts at balance.

It works by obtaining feedback information from the ankle servos. If a force is applied in one of four directions, it will move with the force to avoid falling over. These videos probably show it better.

I've even made the source code available on my blog if you want to have a look.

I'm glad there are people out there who appreciate my hard work! I will try to make some more for you to enjoy.
badcommandorfilename offline
Robot Builder
Robot Builder
Posts: 12
Joined: Fri Dec 11, 2009 11:34 am


5 postsPage 1 of 1
5 postsPage 1 of 1
cron