[ros-users] is a native ROS device possible?
jonathan.bohren at gmail.com
Fri Jun 8 18:37:08 UTC 2012
On Fri, Jun 8, 2012 at 8:06 PM, John Galloway <jrg at toasterfish.com> wrote:
> It appears (i.e. just reading the docs, I haven't brought up a ROS env
> yet) that at present every ROS device has to have an executable (or driver)
> on a host (which is found in the package for that device). It would seem
> ideal that a device using USB (for example) could have its driver embedded
> to allow a generic USB driver to detect that this is a ROS device and link
> it into the graph of nodes (allowing the device to publish and subscribe
> etc) without a device specific driver being needed. Is this possible
> within the current APIs?
There are definitely others who are interested in pushing computation into
embedded systems. At ROSCon last month, Morgan Quigley presented this
related idea, motivated by the needs of realtime control.
You can check out his thoughts on the subject here:
As to having a ROS USB device driver, I don't see any reason why you
couldn't have an embedded system identify itself as a device that talks in
ROS messages, and then have a single driver recognize it via udev rules or
something and hook it up to a given ROS master.
Have you looked at rosserial? http://ros.org/wiki/rosserial While I haven't
used it, it seems like it gets part of the way there by having embedded
devices that talk in ROS messages, and they just connect to a generic
rosserial driver on a host machine: http://ros.org/wiki/rosserial_python
(ahhh it looks like Morgan also replied while I was finishing writing this!)
Dynamical Systems and Control Laboratory
Laboratory for Computational Sensing and Robotics
The Johns Hopkins University
jbo at jhu.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ros-users