>> If we move towards a common driver for ROS, what is the community's
>> opinion on the underlying hardware driver?
> I think it's too early to give a defintive answer here, so the
> discussion of the ROS API and the used underlying hardware driver
> should be independent (at least as long as no new device features are
> found that need to be exposed). It would be best if the implementation
> is geared towards a flexible swap out of the hardware driver.

I would avoid adding an abstraction layer for the hardware driver ; I 
think that such a layer would be overengineering. The interaction with 
the hardware driver means one or two callbacks and about five function 
calls.

Kind regards,

Stéphane

-- 
Dr Stéphane Magnenat
http://stephane.magnenat.net