IEEE Future Directions
Related Articles
I have been invited to give a talk at Fondazione Feltrinelli in Bologna, on November 30th on the interesting aspect of collaboration between humans and robots. Here, and in the next post, I’ll share some ideas, looking forward to your comments that I can socialise at the event.
Humankind has crafted and used tools for as far as we know. Over time these have become ever more sophisticated and, since the XVIII century they have increased its production capabilities making use of the energy provided by steam, first, and electricity, later. This increased production and, most important, decreased the product cost making it affordable to more and more people, thus increasing the overall wellbeing.
In the last 50 years a new tool has changed the scene: the computer. This tool begun by flanking production systems then it became part of the production itself morphing into a blue and white collar worker. Today robots are a growing presence in manufacturing and beyond.
Along this “morphing” robots have acquired, and more so in the coming decade, increasing cognitive capabilities and this is probably what fascinates and concern most the labor world.
Take Baxter, the robot designed and manufactured by Rethink Robotics, an American company founded in 2012 that is producing robots for the manufacturing industry. Baxter has been designed to work in teams staffed with both robots and humans. It is no longer a tool, it has become a co-worker. They gave it (him?) a face – a screen- with eyes able to show emotions to facilitate implicit communications on the job (a communication based on the observation of the environment). Baxter is not “programmed” it learns what it is supposed to do, it is taught by “doing”. You might take its arm and move it to pick up an object, you close its hand to move the object and then you take its other hand, may be in shape of a plier, to fasten a bolt. These guided movements lead to a modification of its knowledge (its software). Actually these modifications are quite complex: as you move its arms and hands video cameras provide it with visual clues and create a context understanding. It can discriminate among objects associating specific actions to specific objects. It is a real process of “learning by doing”.
In these last years Baxter has become smarter (it was improved) and is now capable of learning by observing what its co-workers are doing. It is no longer necessary to guide it by moving its arms and hands to have it learning them, just tell it “Baxter, look how you should do it” and it learns.
There’s more. The latest versions of Baxter are equipped with artificial intelligence software, including deep learning, and are able to conceptualize the knowledge and create new knowledge. Hence they can propose enhancements to the team. It is no longer just a human worker telling Baxter “look how it should be done”, it might as well be Baxter taking the lead and telling a human worker “why don’t you try doing it this way? It might be more efficient!”.
We are approaching a real cooperation among humans and robots on the job.
At the same time we are seeing a variety of robots that are entering into a symbiotic relationship with human workers. This is the case of Ekso, a wearable robot –an exoskeleton- under trial at Ford at their assembly plant in Wayne. This robot relives the worker of the weight he would feel on his arms when he has to work to fasten bolts under the chassis of a car moving over his head on the assembly line. It can manage up to 6.5kg thus relieving the worker from the weight of the wrenches he is using as well as of the weight of his own arms. The trial was supposed to be terminated in August 2017 but the workers felt so good with the robot that requested to keep using it, and they still are.
Ekso has sensors that can detect (guess) the next movements of the arms and steps in to flank the movement relieving it from the weight.
Quite a few research labs are now working on much more sophisticated prototypes, equipped with artificial intelligence and sensors making them aware of their context adapting their behavior in a smart way.
These robots can cooperate in a seamless way with people, both with those working to build the product and with the ones using it (think about the trials of robots by some fast food chains in the US that are cooking hamburgers and lay them on the customer tray once ready).
