The ROS-Industrial ecosystem aims to help develop more specialized industrial robot application software. Gerkey said:
There's only so much you can do with the existing embedded industrial controllers that these companies have spent years developing. Now you can keep that controller box, but add a connection to a PC, for example, with algorithms developed in a research lab.
One big area that could benefit is motion-planning software. When industrial robots are programmed, they must be moved with a joystick from one configuration to the next and programmed to remember each configuration, in a highly tedious and inflexible process. Since the robot doesn't sense where an object is, like the bin it's tossing other objects into, it does all this movement blindly.
In ROS-Industrial, we can add perception and motion-perception software for environments where there's variability in what robots need to do. This can be developed in a university lab, so ROS becomes the pipeline from the lab to the factory. Even if the factory needs to rewrite that software, now they have a new and different capability to use in an industrial environment.
To some industrial robot makers, ROS-Industrial gives the ability to patch in technologies that would otherwise be foreign to them. They can use another software suite that will solve the ability to create increasingly diverse applications, said Erik Nieves, technology director of Yaskawa Motoman Robotics, in an interview. "Where ROS shines is working in unstructured or semi-structured environments where you need more perception, where vision plays a much bigger role, and where that vision requires a more sophisticated response from the robot."
One example of developing for unstructured environments is the 3D mapping software created by the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory to help robots autonomously navigate a constantly changing environment.
Watch videos of robot arms using ROS-Industrial here and here.