- humans use a wide range of dexterity for lots of different tasks (typing on a keyboard, buttoning a shirt, etc.)
- Some tasks require a tool (driving a nail without a hammer is well, shall we say, hard)
- robotic end effectors are typically only good at one kind of dexterity, maybe even better at humans at that one thing, but aren’t great at generalizing. They are effectively hammers, not hands. Hammers are great at driving nails and not much more.
- The best solution to a problem is a function of the available sensing, actuation, computation and communication abilities of the available platform. Usually, there exist trade-offs that allow you to solve a problem using a minimal set of resources but compromise performance characteristics such as speed, accuracy or reliability.
- Robotics problems are different from many problems in pure Artificial Intelligence, particularly those that do not deal with unreliable sensing or actuation.
- The unreliability of sensors, actuators and communication links require a probabilistic notion of the system and the ability to reason with uncertainty.