Seminar: Five hypes and (hence) missed opportunities in robotics
Trends in mainstream robotics
In this seminar, prof. dr. Herman Bruyninckx (Eindhoven University of Technology, KU Leuven) will discuss and illustrate some recent trends in mainstream robotics (academic and industrial alike) that many believe to lead to progress but where he sees only regression with respect to the state of the art.
A (non-exhaustive and non-final) selection of the hypes:
- More data, more computing power, more learning. Problem: intention and abstraction are missing.
- From AI to Explainable AI. Problem: Explainable AI exists for 50 years already, disguised beyond recognition as system identification.
- More sensors, more sensor fusion. Problem: every new “easy” sensor technology deflects necessary funding for real semantic vision.
- More open source means more innovation and better quality. Problem: ROS has become a monopoly, full of fragmentation.
- Fast motion planning brings flexibility and adaptation. Problem: reduction of robots’ control capability to velocity controlled computer games
Bonus: innovation in the Web brings us reactive components and scalability. Problem: ownership of “system state” has disappeared.’
Feel free to bring your own preferred hype to the workshop for discussion!