ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

arm_navigation generation always in collision for Robonaut?

asked 2012-09-04 11:09:15 -0500

shart115 gravatar image

I tried to run the Robonaut2 URDF through the arm_navigation generation tutorial, but it looks like the end-effector is always in collision, and thus i can not generate plans using the planning_components_visualizer.

This is my guess of what's happening based on the end-effector being red upon start up:

image description

and complete inactivity when telling it to "plan" from the menu.

I am using the nasa_r2_common\r2_description\robots\r2c_full_body.urdf.xacro URDF, and choosing the left leg as the 'manipulator' though this seems to work regardless of which of the four limbs i choose.

Is there something i need to check/verify in the advanced wizard to make sure my description is valid?

FYI, I was hoping to push through with the warehouse and motion planning stacks, once I get this working....

Thanks, Steve

edit retag flag offensive close merge delete

2 Answers

Sort by ยป oldest newest most voted
0

answered 2012-09-13 06:21:13 -0500

shart115 gravatar image

I started looking at just the R2 upper body because I had this working with the arm_navigation some months ago, but have been running into similar issues as I have been with the legs.

Just to minimize the chances of collisions, I reduced all the collision models in the entire robot to about 80%, and unchecked all pairs in the advanced mode of the arm nav wizard.

Tabling for now the fact that if i chose more than one kinematic chain, the warehouse viewer wouldn't start properly for me, I'm still getting collisions for all possible movements. The following picture shows what happens as soon as I move the marker (with the robot visualization at a lower alpha). You can see the smaller collision models, none of which are in contact with each other, show up immediately as red (i.e. in collision). This is true for both arms, even when the (only) kinematic chain I have built the models for are between the chest and the left index finger base!

image description

edit flag offensive delete link more

Comments

interestingly, when i remove the collision models for the hands and the first shoulder link, rviz shows the visual models as the collision models when i start up the warehouse viewer. that cant be right....

shart115 gravatar image shart115  ( 2012-09-13 07:53:03 -0500 )edit
0

answered 2012-09-05 05:58:38 -0500

Steve,

This has occured to me when there are two non-adjacent links that are close to each other. They may be collision free in reality, but the coarse collision models often aren't. On industrial arms, this happens in the wrist quite often. If this is the case, you should be able to disable collision checking (assuming that this is safe).

FYI...I've created a walk through for industrial arms ( http://ros.org/wiki/Industrial/Tutorials/Create_a_Arm_Nav_Pkg_for_an_Industrial_Robot ). This may have information not in the standard tutorial.

edit flag offensive delete link more

Comments

Thanks, for the response. Using the advanced mode, In most of the collision checks, links show up---despite the ROS-Industrial tutorial blurps often saying that there "should be no reason why this might be the case."

shart115 gravatar image shart115  ( 2012-09-05 07:37:02 -0500 )edit

is the best course of action to go through the URDF to see how many of these collisions I can get rid of?

shart115 gravatar image shart115  ( 2012-09-05 07:37:32 -0500 )edit

Might not be the best way, but at least I'm doing that quite regularly after the initial set-up with the configuration wizard. Even when using the normal mode, I often have to remove some links. Remember that collision checking is using an additional padding on top of your robot's collision model.

bit-pirate gravatar image bit-pirate  ( 2012-09-09 13:11:20 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2012-09-04 11:09:15 -0500

Seen: 329 times

Last updated: Sep 13 '12