Predictions and Problems please....
Predictions and Problems please....
by NovaOne
Sun Sep 16, 2007 9:07 pm
Everyone has different motivations for building, modifying, and programming their bipeds and contributing to this site, maybe some peoples motivations are selfish, but I like to think most of us have similar goals? However being a quite limited electro-chemical humanoid, I am finding it difficult to see the forest for the trees.
Bill Gates has said he thinks that the robot industry is akin to the PC industry 30 years ago ...I would like to think he is right. To summarize, most of the key players are in place, but they are making completely incompatible machines. It seems like just round the corner is the open architecture Personal Robot platform that, (just as the IBM PC gave birth to a menagerie of companies, that designed and supplied IBM-compatible PC hardware), will make possible a useful and self sustaining Personal Robot industry? What will the killer app's be?......I've gotten carried away again, ignore this paragraph.
Coming back down to earth:
How to you guys view the Biped road map ahead in the near future, without wild speculation?
Being realistic what are the problems that currently, really need solving?
Organizations like Sony and Honda and some universities seem to have solved some of the biggies (ie their robots are very impressive at least), but are they scalable? Or are they even on the right track?
Chris
Bill Gates has said he thinks that the robot industry is akin to the PC industry 30 years ago ...I would like to think he is right. To summarize, most of the key players are in place, but they are making completely incompatible machines. It seems like just round the corner is the open architecture Personal Robot platform that, (just as the IBM PC gave birth to a menagerie of companies, that designed and supplied IBM-compatible PC hardware), will make possible a useful and self sustaining Personal Robot industry? What will the killer app's be?......I've gotten carried away again, ignore this paragraph.
Coming back down to earth:
How to you guys view the Biped road map ahead in the near future, without wild speculation?
Being realistic what are the problems that currently, really need solving?
Organizations like Sony and Honda and some universities seem to have solved some of the biggies (ie their robots are very impressive at least), but are they scalable? Or are they even on the right track?
Chris
by Robo1
Sun Sep 16, 2007 10:11 pm
The real problem is "safe critical machines". e.g. how can you guarantee that if you robot breaks it will not kill some one.
There is a project at are Uni trying to start to address this problem. Unlike planes where you can have 3 different types of system so there all independent, you can have the same thing will a biped as it's brain will always be electronic.
to give you an idea of what faces us/"the industry" can you say for certain what will happed if pin 134 on the control board breaks what will be the outcome is if safe. then you get more and more of these pin 3 and pin 127. think about how many of these controllers there are! you don't want you biped to break then kill someone, I todays climate you will probably end up in prison.
Bren
There is a project at are Uni trying to start to address this problem. Unlike planes where you can have 3 different types of system so there all independent, you can have the same thing will a biped as it's brain will always be electronic.
to give you an idea of what faces us/"the industry" can you say for certain what will happed if pin 134 on the control board breaks what will be the outcome is if safe. then you get more and more of these pin 3 and pin 127. think about how many of these controllers there are! you don't want you biped to break then kill someone, I todays climate you will probably end up in prison.
Bren
by NovaOne
Sun Sep 16, 2007 10:46 pm
I'm quite "conservative" and believe a really useful personal biped may be more 30 years off. I was dreaming, in that paragraph of my post.
I can't even imagine a complex robot with so may safety trips. It would be shutting down all the time reporting all manner of errors. -Bren, what a nightmare vision you have!
What about the problems we will be seeing in the near future, say next five years
(with robots no bigger than 2 feet high)?
Chris
I can't even imagine a complex robot with so may safety trips. It would be shutting down all the time reporting all manner of errors. -Bren, what a nightmare vision you have!
What about the problems we will be seeing in the near future, say next five years
(with robots no bigger than 2 feet high)?
Chris
by BillB
Mon Sep 17, 2007 11:16 am
The obvious point is that we still need to find a practical use that Bipeds can do much better than wheeled devices. Bipeds are not going to vacuum a floor better than a wheel device in th next 5 years and any mainstream 'entertainment' novelty wears off pretty quickly. Anyway that is a problem for beyond 5 years.
The 5 year problem will be to make it possible to add a couple of servos to the biped without having to spend an eternity reprogramming all the moves because the new servo has subtly changed the centre of mass and effectivly invalidated all the preprogrammed moves.
What would help is a standard notaion/language that describes the target position of he limbs and lets the robot do the rest. If a new servo or mass is added then it should not effect the target positions of the limbs, and the robot software can calculate how the servos should be positioned.
The 5 year problem will be to make it possible to add a couple of servos to the biped without having to spend an eternity reprogramming all the moves because the new servo has subtly changed the centre of mass and effectivly invalidated all the preprogrammed moves.
What would help is a standard notaion/language that describes the target position of he limbs and lets the robot do the rest. If a new servo or mass is added then it should not effect the target positions of the limbs, and the robot software can calculate how the servos should be positioned.
by JonHylands
Mon Sep 17, 2007 1:52 pm
The problem with pre-calculating where all the servos are supposed to go is the robot is then incapable of responding to changes in the outside environment - the ground slope changes, someone or something pushes the robot, etc. What bipeds really need is a concept of balance, and a mechanism that allows the robot to always try and stay in balance, and to be able to do that while moving around.
Once you have that, then adding weight here and there on the robot won't make any difference.
- Jon
Once you have that, then adding weight here and there on the robot won't make any difference.
- Jon
by NovaOne
Mon Sep 17, 2007 8:55 pm
BillB:
I like the sound of a standard movement/limb position language! I imagine the standard movement protocol would have several layers?
I agree adding extra DOF is a real headache at the moment. Wouldn't a physics engine be needed for the software to automatically adjust/re-map how the servos should be positioned?
I suppose currently this would need integrating into the development environment? But could it be done one board when required, (ie if a limb goes faulty)? If enough processing power became available?
JonHylands:
Do I need to ask, Jon, are you working with a Kalman-filters, to combine the gyroscope and accelerometer values?
If so are you able to obtain useful output values from the filter that can be processed fast enough to be useful during the walking phase?
What would help is a standard notaion/language that describes the target position of he limbs and lets the robot do the rest. If a new servo or mass is added then it should not effect the target positions of the limbs, and the robot software can calculate how the servos should be positioned.
I like the sound of a standard movement/limb position language! I imagine the standard movement protocol would have several layers?
I agree adding extra DOF is a real headache at the moment. Wouldn't a physics engine be needed for the software to automatically adjust/re-map how the servos should be positioned?
I suppose currently this would need integrating into the development environment? But could it be done one board when required, (ie if a limb goes faulty)? If enough processing power became available?
JonHylands:
What bipeds really need is a concept of balance, and a mechanism that allows the robot to always try and stay in balance, and to be able to do that while moving around.
Do I need to ask, Jon, are you working with a Kalman-filters, to combine the gyroscope and accelerometer values?
If so are you able to obtain useful output values from the filter that can be processed fast enough to be useful during the walking phase?
by JonHylands
Mon Sep 17, 2007 9:47 pm
I've looked at using a kalman filter, but haven't gotten to that point yet.
You shouldn't have any trouble getting values fast enough using my IMU - I update the values from the sensors at 100 Hz, and you can retrieve the values that fast from the IMU if you want. Given that the processing is happening on a PC, you've got enormous amounts of processing power available to use this sensor information.
- Jon
You shouldn't have any trouble getting values fast enough using my IMU - I update the values from the sensors at 100 Hz, and you can retrieve the values that fast from the IMU if you want. Given that the processing is happening on a PC, you've got enormous amounts of processing power available to use this sensor information.
- Jon
)
by NovaOne
Tue Sep 18, 2007 8:49 pm
I would like to think that the future of biped robots is open source, but then I remembered years ago (2002) coming across, the open pino platform :
http://www.symbio.jst.go.jp/PINO/
I am probably wrong, but it seems to have (inspired) given birth to smaller robots like bioloids, robonovas, khrs etc, then just gone stagnant and died?
The last dieing breaths seemed to be the morph 3 then, the ZMP nuvo....? Last time i looked nuvo cost £4000, Surely no one bought it?
Were these all dead branches in the biped development tree?
How much of PINO is in the robots being developed by "Savvy Roboteer's"
Chris
http://www.symbio.jst.go.jp/PINO/
I am probably wrong, but it seems to have (inspired) given birth to smaller robots like bioloids, robonovas, khrs etc, then just gone stagnant and died?
The last dieing breaths seemed to be the morph 3 then, the ZMP nuvo....? Last time i looked nuvo cost £4000, Surely no one bought it?
Were these all dead branches in the biped development tree?
How much of PINO is in the robots being developed by "Savvy Roboteer's"
Chris
by Humanoido
Mon Sep 24, 2007 3:06 pm
There's certainly a revival going on with 2 servo bipeds - like the amazing two servo biped that rock dances like there's no tomorrow! So far, new programs for this tiny bot have soared in the past couple months. (open source and free) It looks like its a lot of fun to program and cute is off the scale!
It dances, it even roller skates. The latest is a soccer playing program, and a program for wireless.
humanoido
PENGUIN SOCCER
http://www.youtube.com/watch?v=5Duxpxu4Bhs
PENGUIN HAPPY DANCE
http://www.youtube.com/watch?v=9owSIXhh_Mo
PENGUIN DOES THE ROBOT
http://www.youtube.com/watch?v=LpH3ZsNwBn8
PENGUIN ROLLER SKATE DANCE
http://www.youtube.com/watch?v=G_H3teQdxxw
PENGUIN PRODUCT PAGE
http://www.parallax.com/detail.asp?product_id=27313
PENGUIN DOCUMENTATION
http://www.parallax.com/dl/docs/prod/robo/27313-6%20PenguinDoc-v1.0.pdf
PRS PENGUIN ROBOT SOCIETY
penguin dot robot at yahoo dot com
PENGUIN CODE
http://forums.parallax.com/forums/default.aspx?f=10
http://www.parallax.com/detail.asp?product_id=27313
It dances, it even roller skates. The latest is a soccer playing program, and a program for wireless.
humanoido
PENGUIN SOCCER
http://www.youtube.com/watch?v=5Duxpxu4Bhs
PENGUIN HAPPY DANCE
http://www.youtube.com/watch?v=9owSIXhh_Mo
PENGUIN DOES THE ROBOT
http://www.youtube.com/watch?v=LpH3ZsNwBn8
PENGUIN ROLLER SKATE DANCE
http://www.youtube.com/watch?v=G_H3teQdxxw
PENGUIN PRODUCT PAGE
http://www.parallax.com/detail.asp?product_id=27313
PENGUIN DOCUMENTATION
http://www.parallax.com/dl/docs/prod/robo/27313-6%20PenguinDoc-v1.0.pdf
PRS PENGUIN ROBOT SOCIETY
penguin dot robot at yahoo dot com
PENGUIN CODE
http://forums.parallax.com/forums/default.aspx?f=10
http://www.parallax.com/detail.asp?product_id=27313
Last edited by Humanoido on Tue Oct 16, 2007 4:48 am, edited 1 time in total.
