This page has a rather grand title, but will just deal with my opinions about the current focuses on robotics trends. I think humanoid robotics is flashy, but might not be the best approach. A lot of time is being spent on walking and balancing robots, but what is that actually useful for? There’s a lack of imagination in form and hardware design here.

Robots break down into a few things

  1. robots meant for a specific purpose and meant to do that purpose as reliably, repeatably, and rapidly as possible
  2. robots meant for general purpose tasks and meant to be exposed to many different environments and conditions/tasks

The pareto frontier for robots feels somewhat tight - this is where we differ from llms in general. Getting more general does not make us better at specific tasks.

I think there needs to be some sort of state transition in robotics for us to reach a point where generalizing makes us better at specific tasks. We haven’t had our “chatgpt moment” yet.