Home | Tutorials | Wiki | Issues
Ask Your Question

Is it possible to distribute sensors over multiple computers?

asked 2012-11-28 04:37:08 -0500

Max Pfingsthorn gravatar image

I would like to simulate a number of robots (up to let's say 10) in real time, each with a significant sensor suite. Is it possible to distribute sensors as slaves to the physics simulation across multiple machines? If it's not right now, does the architecture readily support it? I could imagine a plugin to the main server to broadcast world changes, and a "fake" physics that passes that on to the local sensors on each slave machine.

edit retag flag offensive close merge delete

2 Answers

Sort by ยป oldest newest most voted

answered 2013-01-11 19:47:55 -0500

nkoenig gravatar image

Sensor simulation is currently not distributed. I've created a ticket for this: https://bitbucket.org/osrf/gazebo/issue/383/distributed-sensors

edit flag offensive delete link more

answered 2012-12-04 15:51:34 -0500

SL Remy gravatar image

You won't win any awards for what I'm about to suggest, but it does provide a way :)

It's not clear what type of sensors you're intending, but... if you have a sensor suite that is based on the pose of each robot, then you could via the standard magic have multiple machines that subscribe to the odom or the simulation states of the robots that you need, perform the processing in their local environment (i.e. on that machine) and then publish so any node that needs to have sensor information from that machine can have it (since they're all connected to the same ROS master)..

edit flag offensive delete link more


The problem here is that those sensors that only need the pose (e.g. IMU) are really not computationally intensive. Range sensors on the other hand are, but they need the world geometry.

Max Pfingsthorn gravatar imageMax Pfingsthorn ( 2013-01-24 07:26:32 -0500 )edit
Login/Signup to Answer

Question Tools

1 follower


Asked: 2012-11-28 04:37:08 -0500

Seen: 370 times

Last updated: Jan 11 '13