miltos's profile - activity

Hello!

I got this message: BUG: Got a transition to CommState [PENDING] when our in SimpleGoalState [ACTIVE]

This seems to be printed in the simple_action_client.h when the client receives a state different that what it expects. How is this possible? What did I do wrong?

Thanks

While streaming with theora image_transport, I receive the following error:

a message of over a gigabyte was predicted in tcpros. that seems highly unlikely, so I'll assume protocol synchronization is lost.

What might cause this error?

I have a set of launch files that each of them initiate a subsystem. However, some of the .launch files contain dependencies on a subset of the system. I could include all dependencies in a .launch file and include it in all other launch files. However, for running the complete system, I include all .launch files in a "global" launcher. This causes the common nodes to start multiple times. How could I solve that? Is there any way to "declare once" some nodes?

I am sending a tf transform every 3s. What happens when I try to transform a point with timestamp between these 3 seconds? Which transform is used? Does an interpolation happen?

Thanks

Hi! I am trying to synchronize a camera input with an servo with no feedback and the tf library. More specifically:

1) There is a camera (cameraNode) that asynchronously (with a ros::timer) takes pictures and posts them at a ROS topic with a timestamp.

2) There is a node (servoNode) that controls a servo and some other hardware, that has no feedback.

3) There is a node (armNode) that controls a set of motors (and the servo through servoNode) and publishers tf transforms with their positions.

4) There is a node (visionNode) that processes the images and publishers results about the images (with the image timestamp).

5) There is a node (dataNode) that receives the results and time-travels (with tf) to obtain actual points in a coordinate system.

All these seem to work fine individually, but there are times that the camera takes a picture just before the servoNode returns and the armNode publishes the new tf transformation. Then the dataNode when it uses the lookupTransform (with the camera timestamp) gets a transformation that is not correct. How could these be synchronized? Should the armNode stop broadcasting tf transforms (and if yes, how would that change the dataNode's ability for correct transformation)?

Using a tf::MessageFilter doesn't seem right because tf transforms might be sent just before the servos started moving and the camera might take up to 200ms to take a picture. Any suggestions?

no activity > 1 month, closing

As the error message indicates, it means that the connection some how became out-of-sync. This could potentially happen if streaming between multiple computers and somehow the libraries are not the same, or something even more esoteric.

One way would be to give each of your subsystem launch files an argument (which could default to true) and use the if attribute within those launch files to start your common nodes only when that argument is set to true. Then, in your global launcher, set the same argument to false when you include the subsystem launch files and launch the common nodes once.