Difference between revisions of "VVV09 Kinematic, my friend"

From Wiki for iCub and Friends
Jump to: navigation, search
 
Line 1: Line 1:
VVV09 Kinematic, my friend.
+
== VVV09 Kinematic, my friend ==
  
 
The aim subject is the learning of the forward kinematic model.
 
The aim subject is the learning of the forward kinematic model.
  
 
Why learn the model, we already have an analytic method ?  
 
Why learn the model, we already have an analytic method ?  
- A human like approach.
+
* A human like approach.
- To validate the feasibility of such kind of learning.
+
* To validate the feasibility of such kind of learning.
  
 
Why the forward model ?
 
Why the forward model ?
- To keep all the information of the forward model (i.e infinity of pseudo-inverse, null space computation).
+
* To keep all the information of the model (i.e infinity of pseudo-inverse of the Jacobian, null space computation).
  
 
Why the kinematic ?
 
Why the kinematic ?
- Because the geometric model is not easily invertible.
+
* Because the geometric model is not easily invertible.
  
 
What kind of algorithm is using ?
 
What kind of algorithm is using ?
- LWPR, (Locally Weighted Projection Regression) [Vijayakumar 2005]. A powerfull online learning algorithm based on multiple linear regression computation in the space the most adapted (PLS).
+
* LWPR, (Locally Weighted Projection Regression) [Vijayakumar 2005]. A powerfull online learning algorithm based on multiple linear regression computation in the space the most adapted (PLS).
  
 
What are the results ?
 
What are the results ?
- The model is learned and use "successfully" on the robot.
+
* The model is learned and uses "successfully" on the robot.
  
 
What about the "keep all the information" ?
 
What about the "keep all the information" ?
- With the model learned we are able to control two end effector with a priority on one of the two.
+
* With the learned model we are able to control two end-effectors with a priority on one of them.
  
 
So, what did you do here ?
 
So, what did you do here ?
- I tried to learn the model ... without the model ... -> Vision.
+
* I tried to learn the model ... without the model but with the vision.
  
 
And ?
 
And ?
- I have learn the model by attended motor babbling (As we play with the "end effector of a Baby").
+
* I have learned the model by attended motor babbling (As we play with the "end effector" of a baby).
- I "wanted" to be able to track two objects (my end effector and another object ... But it is apparently not so easy :p) in order to complete the learning of the model still as if we were playing with a child.
+
* I "wanted" to be able to track two objects (my end effector and another object ... But it is apparently not so easy :p) in order to complete the learning of the model still as if we were playing with a child.

Revision as of 10:49, 29 July 2009

VVV09 Kinematic, my friend

The aim subject is the learning of the forward kinematic model.

Why learn the model, we already have an analytic method ?

  • A human like approach.
  • To validate the feasibility of such kind of learning.

Why the forward model ?

  • To keep all the information of the model (i.e infinity of pseudo-inverse of the Jacobian, null space computation).

Why the kinematic ?

  • Because the geometric model is not easily invertible.

What kind of algorithm is using ?

  • LWPR, (Locally Weighted Projection Regression) [Vijayakumar 2005]. A powerfull online learning algorithm based on multiple linear regression computation in the space the most adapted (PLS).

What are the results ?

  • The model is learned and uses "successfully" on the robot.

What about the "keep all the information" ?

  • With the learned model we are able to control two end-effectors with a priority on one of them.

So, what did you do here ?

  • I tried to learn the model ... without the model but with the vision.

And ?

  • I have learned the model by attended motor babbling (As we play with the "end effector" of a baby).
  • I "wanted" to be able to track two objects (my end effector and another object ... But it is apparently not so easy :p) in order to complete the learning of the model still as if we were playing with a child.