Put a Spring in your Step
Put a Spring in your Step
by Dewey
Sun Dec 12, 2010 4:07 pm
First rule of Bot building - 'you never have enough servo torque!'
Often the situation is made worse by the fact that some joint servos may be exerting a large torque just to handle the effects of mass/gravity. I've been experimenting (just on a loaded test arm) of using elastic force to 'cancel' or reduce static loads, actually I'm using nothing more than elastic bands.
Here's the idea:
Suppose a servo can exert torque in the range -10 units to +10 units. A 'gravity loaded servo is at rest but is exerting say 6 units of torque.
If we wish to 'move' against this we only have 4 units available, move with it we have 16 units. If we employ elastic tension to 'cancel' the +6, (a tensioned elastic band) then the servo has available -10 and +10 torque range available again.
This simple arrangement has limits as we move we eithe reduce or increase the 'help' the elastic gives us, but being elastic we do recover this energy when we reverse the direction.
Arranging the elastic 'helpers' such they don't interfere or get caught up is also an issue.
Nevertheless this spring counterbalancing (the angle poise lamp idea) give impressive gains when working against gravity.
Anyone else been down this road, or is this all to simplistic an idea and I'm missing something fundamental.
Dewey
Often the situation is made worse by the fact that some joint servos may be exerting a large torque just to handle the effects of mass/gravity. I've been experimenting (just on a loaded test arm) of using elastic force to 'cancel' or reduce static loads, actually I'm using nothing more than elastic bands.
Here's the idea:
Suppose a servo can exert torque in the range -10 units to +10 units. A 'gravity loaded servo is at rest but is exerting say 6 units of torque.
If we wish to 'move' against this we only have 4 units available, move with it we have 16 units. If we employ elastic tension to 'cancel' the +6, (a tensioned elastic band) then the servo has available -10 and +10 torque range available again.
This simple arrangement has limits as we move we eithe reduce or increase the 'help' the elastic gives us, but being elastic we do recover this energy when we reverse the direction.
Arranging the elastic 'helpers' such they don't interfere or get caught up is also an issue.
Nevertheless this spring counterbalancing (the angle poise lamp idea) give impressive gains when working against gravity.
Anyone else been down this road, or is this all to simplistic an idea and I'm missing something fundamental.
Dewey
by i-Bot
Sun Dec 12, 2010 5:35 pm
I played around a bit with elastic linkages looking at variable compliance. Basically like have in push rods between the servo and the actual pivot, except the push rods were elastic (o ring rubber) pull rods. I had a magnetic encoder on the actual pivot. I made some easy catch and play moves a bit like a bendy toy, but also was able to bounce. No help with static loads though, I never tried it with one rod to a fixed anchor point.
Take a look at some of the stuff Limor is doing to build elastic behavior into the servo.
Maybe Limor should organise a webcast or a chat to present his work and get feedback on similar projects like yours.
Take a look at some of the stuff Limor is doing to build elastic behavior into the servo.
Maybe Limor should organise a webcast or a chat to present his work and get feedback on similar projects like yours.
by Fraser
Sun Dec 12, 2010 6:14 pm
I recently saw a bioloid humanoid with rubber bands stretched over its shoulders, I can't remember what thread, perhaps it was your work?
A clever idea to get near zero holding torques on a standing robot, cheaper than servo upgrades and should extend battery life.
Might be worth developing further with thin silicone tubes and custom brackets, small rollers on knees, shoulders etc. to prevent chaffing.
Animal running has evolved energy storage and recovery with tendon elasticity, in the future it might be essential tech for fastest running bots.
A clever idea to get near zero holding torques on a standing robot, cheaper than servo upgrades and should extend battery life.
Might be worth developing further with thin silicone tubes and custom brackets, small rollers on knees, shoulders etc. to prevent chaffing.
Animal running has evolved energy storage and recovery with tendon elasticity, in the future it might be essential tech for fastest running bots.
by Fraser
Sun Dec 12, 2010 6:50 pm
Robotic leg with mechanical springs for energy storage and recovery.
Different to your idea, I think they are not driving the skeleton directly but through the springs, similar to what i-Bot tried, driving through elastic.
A humanoid with legs like that would be a fearsome sight!
Different to your idea, I think they are not driving the skeleton directly but through the springs, similar to what i-Bot tried, driving through elastic.
phpBB [media]
A humanoid with legs like that would be a fearsome sight!
by garycmartin
Mon Dec 13, 2010 4:42 am
Fraser wrote:I recently saw a bioloid humanoid with rubber bands stretched over its shoulders, I can't remember what thread.
Not sure if this was the project you had in mind, but your comment reminded me of the rather lovely Acroban Humanoid Project:
http://flowers.inria.fr/acroban.php
It's a custom build using Robotis RX-28 and RX-64 motors, coupled with springs and elastic.
by Fraser
Mon Dec 13, 2010 1:26 pm
garycmartin wrote:Fraser wrote:I recently saw a bioloid humanoid with rubber bands stretched over its shoulders, I can't remember what thread.
Not sure if this was the project you had in mind, but your comment reminded me of the rather lovely Acroban Humanoid Project:
http://flowers.inria.fr/acroban.php
It's a custom build using Robotis RX-28 and RX-64 motors, coupled with springs and elastic.
Hi, I hadn't seen that before, neat video, the kids were fascinated with him. Very cool to see the servos actively driven to appear compliant and floppy.
by taiowa
Sat Apr 02, 2011 8:52 pm
I have tested using ribbons of rubber cut from elastic gym rubber bands to counterweight the robot weight on the foot joints. I taped the rubber bands to the bottom of the hip servos, routed the band over the front of the knee joint, and then taped the other end to the back of the ankle joint.
With the rubber bands in place the robot was able to stand up much faster from sitting position and also to walk faster.
The design was so easy I have been wondering why hasn't anyone else come up with the idea. My batteries went dead quite soon after that so my robotis has been collecting dust for quite a long time and I haven't been able to develop the idea further.
With the rubber bands in place the robot was able to stand up much faster from sitting position and also to walk faster.
The design was so easy I have been wondering why hasn't anyone else come up with the idea. My batteries went dead quite soon after that so my robotis has been collecting dust for quite a long time and I haven't been able to develop the idea further.
by limor
Sun Jun 05, 2011 8:54 pm
taiowa wrote:I have tested using ribbons of rubber cut from elastic gym rubber bands to counterweight the robot weight on the foot joints. I taped the rubber bands to the bottom of the hip servos, routed the band over the front of the knee joint, and then taped the other end to the back of the ankle joint.
With the rubber bands in place the robot was able to stand up much faster from sitting position and also to walk faster.
The design was so easy I have been wondering why hasn't anyone else come up with the idea. My batteries went dead quite soon after that so my robotis has been collecting dust for quite a long time and I haven't been able to develop the idea further.
pictures / video please
