Tools to Teammates
Back in September I had the honor of delivering the keynote at ASME’s Conference on Robotics for Inspection and Maintenance in College Station, TX. It was a great conference, full of customers, technologists, and investors with a shared passion for the topic at hand.
The focus of my talk was on the journey we at GE Research have embarked on to build intelligent systems that work in complex environments to address dull, dirty, and dangerous tasks; be it war-fighting, confined space inspection and repair, reduced crew operations, or any of a host of industrial field service tasks.
In my talk, I presented 3 phases we see things moving through in the coming years. In each phase, robots/machines/vehicles increase the complexity of tasks they can handle and the level of autonomy with which they handle them. Below is a way in which we view the overall space and where it’s all headed. You can think of the space of intelligent machines as being governed by 2 dimensions, “Level of autonomy/automation” and “Task complexity”. If you plot the state of the art in these two dimensions, you‘ll notice there’s a frontier that becomes apparent with two poles: 1. Systems that do extremely complex tasks (e.g. surgery) that are directly controlled by humans, and 2. Systems that do relatively low complexity tasks that are completely automated (e.g. a basic pick-and-place task). Everyone, whether you’re an automotive company building an autonomous car, or an industrial inspection company offering robotic enabled inspection, wants to move up and to the right; as you automate more and more complex tasks, the value proposition of the system increases. Advances in core technology that we focus on including perception, reasoning, dexterity, and trust, will enable this shift.
So, let’s briefly discuss the 3 phases…
Phase 1 – robots as tools – these are primarily human controlled systems that allow us to do things we couldn’t otherwise; examples would be systems like minimally invasive surgical tools, the Davinci robot, the drone you use to take pictures of your rooftop, and even the car you drive today. We are building tools that allow field engineers to rapidly inspect critical infrastructure and machinery including turbines, aircraft engines, pipes, and planes.
Phase 2 - robots as teammates – here more intelligence is added to the robot device that enables collaboration with the human – think R2D2, Rosie the Robot, and HAL. We are building a number of intelligent systems that learn from humans and collaborate to enable manned-unmanned teaming (MUM-T) in both military and industrial use cases.
Phase 3 – self-sustaining systems – these are systems that operate on complex tasks without human intervention, they live in their target environment and humans only get involved to handle rare errors and exceptions – think Wall-E. We are just starting to scratch the surface of what’s possible here. More to come.
In the coming weeks and months, I’ll be sharing specific examples (hopefully with videos) of the various tools and teammates we’re building for the world. Stay tuned.
Looking forward to it John. Great seeing you in College Station as well