This past week, I attended the annual RoboBusiness conference in Santa Clara, sponsored by Robotics Business Review. The conference featured a number of keynote addresses related to recent advances in AI and grasping technology, as well as five different technical tracks focused on the robotics industry and aspects of robot design. There was also a moderately sized hall with over 70 industry exhibitors that represented all parts of the robotics ecosystem.
During the conference, I met with representatives from over three dozen companies involved in all aspects of the robotics industry: robot suppliers, component suppliers, new technology developers, designers, and consultants, as well as with press and public relations firms.
Some key takeaways from the conference in general:
- Artificial Intelligence (AI), and so-called “deep learning” – multi-layer neural networks with millions of training objects – is emerging from the laboratory and entering the mainstream. Deepu Talla of NVidia delivered the first conference keynote presentation on this topic. Peter Abeel of Covariant (also a Stanford professor) shared some additional implementation examples, as did Priyanka Bagade of Intel, in their respective Robot Design and Engineering Forum talks.
- AI is being applied not only to things like vision and image recognition systems but also to other areas such as grasping technology. Ken Goldberg of the University of California at Berkeley described the “New Wave in Robot Grasping” in his keynote talk, with Google’s “arm farm” as a backdrop. He described new advances arising from the application of deep learning to an ABB Yumi robot grasper. Employing a database of almost 7 million objects, the accuracy and speed of the system approaches human levels.
- Robotics has now become so endemic to the workplace that multiple panel discussions and workshops were devoted to the topic of social implications. I had one very animated discussion on this topic with Dr. Joanna Korman, a social cognitive science with the MITRE Corporation, at a conference networking event.
One takeaway is somewhat cataphatic: There were no discussions of any significant advances in robot transmission technology.
While there were some new product announcements on the show floor, what was presented in the technical fora was essentially technology that has been around since the 1950’s or earlier. One-on-one discussions with the robot and joint developers revealed deep dissatisfaction with the weight, efficiency and other characteristics of harmonic drives, but there’s a sort of resignation to the technology since nothing new – until now, with Motus – has come forward.