[Robotics] Origami-folding robot

[Benoit Bolsee]

-----Original Message-----
From: robotics-bounces at blender.org [mailto:robotics-bounces at blender.org]
On Behalf Of Andy Mills
Sent: mardi 5 janvier 2010 21:52
To: robotics at blender.org
Subject: [Robotics] Origami-folding robot


Hello,

I've decided to use Blender to model a robot/machine idea I have, mostly
because it already has the physics engine built in and the soft-body
objects look promising.

At this point, I have a model of what I envision the robot to be like
(some things are likely to change). However, I have some quetions:

- It's not that important right now, but as it stands, the robot is
purely static objects, so no collsion. I notice that if I apply physics
to just one part, that part tends to fly away. Is it possible to build
an armature that's connected to a "static body" but the armature
controls a rigid body object? Will I have to create the whole thing as
rigid body? (There are some pieces that are stationary that I imagine
could be static.
*Note, I don't think it's so necessary to have this answered by any of
you. I have a feeling I can find a tutorial or relevant info to this,
but any help would be appreciated.
 

I'm not sure what you mean by static/dynamic object. The normal way to
do a robot is to set an armature and parent a series of object to each
bone to represent the segment of the robot arms. These objects must be
static because a rigid body cannot be the child of anything. Static
object doesn't mean no collision. The robotic arm can interact with
rigid bodies in the scene but it will not get force feedback. To move
the robot, set the armature pose, either automatically with IK or
manually by setting the bone pose directly. There is an API in Blender
2.5 that simplifies the control of the bones by allowing to set the
joint angle (in Blender 2.49 you must set the bone matrix).

 
- Do any of you have experience attempting to simulate paper, and
manipulate it? Do you think the soft bodies will be satisfactory? I'm
really wondering if the soft bodies can "keep a fold", like if you were
to take a piece of paper, fold it in half, then set it on a table with
the fold up, like ^ whether it would maintain that fold, but slowly lose
it.
I think I might end up being alone on this one...
 

Unfortunately, Bullet Soft Body are not so useful. It is not possible to
have a folding effect, it is also very difficult to grab a soft body:
friction and plastic deformation are not properly implemented. Collision
is also poorly implemented: soft bodies tend to 'sink' through static
object. You may try to experiment with the various parameters but my own
experiments have been quite deceiving. 

 
- Another one I don't necessarily NEED help on, but any insight might
help. The game seems to play just fine from Blender, but when I try to
export a binary, it seems as if the camera is facing away from the
"scene." I haven't tried to set up controls for the camera, yet, but I'm
hoping that doing so will allow me to maneuver/focus the camera on the
scene (although I feel like it should be oriented as set up in the
scene/3D view. (I'm not sure if the camera is facing away or if the
objects aren't being drawn). Have any of you experienced this?
 

When running the game from Blender, any view point you have in the 3D
view will be converted in a temporary camera in the game. When running
the game from the binary, you can only use the real camera. To test the
camera in Blender, press 0 in the 3D view before running the game: this
will put you in the camera position and you will see the scene from the
camera view point. Fix its position if it doesn't point to the right
position.

 
 
- As it stands, I'm not currently using any sensors. At a minimum, I'm
expecting to use some IR sensors to ensure the paper's alignment, but
I'm hoping I won't need very sophisticated sensor systems. I've been
looking at Pyro and Sirbot to satisfy sensor simulation. What are you
using?

I'm willing to share the .blend file with you to get a better idea of
what I'm talking about. Currently, the robot is controllably only
manually via keyboard.

Thanks in advance. I hope not to be bothersome. If my project offers an
opportunity, I hope to be able to contribute to the robotics project.

I'm striving to "complete" a thorough simulation within the next six
months.

Regards,

Andy Mills


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/robotics/attachments/20100106/3b03c8f8/attachment.htm