Revision history [back]

ROS-I brings the rich world of developed solutions (ROS packages) and the ones in research to be integrated to the industrial robots (manipulators), the main issue that ROS has is a lack of reliability and safety concern. Those are basically the main points in which ROS-I is working together with the robot vendors, industry can't afford having a robot that suddenly behave in an unexpected manner. Pros of ROS-I:

Let's developers create solutions that are cross robot, so no matter the brand, the same solution will run for any ROS-I compatible robot.
You can integrate the robot capabilities to existing solutions (ROS packages) that enhance the capabilities and functionalities of the robot in the production line: Like object recognition, point cloud, et c.
Brings the open-source way of development (community) to the industrial robots, letting industries be less dependent on a specific vendor.
Programming a tasks becomes totally independent of the vendor specific OS.

Cons of ROS-I:

It is not reliable yet.
takes too much time to prepare and perform the same high level movements (like welding) that a robot can perform while being easily program with its own vendor programming interface.
Industry requires a reliable and precise robot that can perform the same movement more than 10.000 times without ever having weird behaviours. So far, MoveIt is not able to guarantee that this won't happen, Descartes gives some hopes in this aspect, giving to the user more control in the way that the robot must move, but is too clumsy to use, so, for creating a task, it is not as fast as using the vendor way of programming the robot.

At the end is a matter of a few moths (I hope) to make ROS-I reliable, user friendly and safe enough to take all vendor OS out of the market. (I'm also working on this, look at: https://github.com/fontysrobotics/robot_pick_place)