Robotics is the engineering and science of machines that perceive their surroundings, decide what to do, and physically act - the loop of sensing, planning, and actuation. It overlaps with artificial intelligence but is not the same thing. AI can live entirely inside a computer, manipulating text or numbers. A robot has a body, so it must contend with friction, gravity, noisy sensors, imperfect motors, and a world that does not pause while it thinks.
That embodiment is why robotics has repeatedly lagged behind software AI. The pattern is captured by Moravec’s paradox, named for roboticist Hans Moravec: tasks that feel intellectually hard for humans, like chess or arithmetic, turn out to be relatively easy to program, while tasks that feel effortless, like walking across a cluttered room or recognizing a face, are extraordinarily hard for machines. In his own writing, Moravec noted that for decades the computing power in advanced AI and robotics systems was “stuck at insect brainpower of 1 MIPS,” and only when hardware leapt forward did machines start “reading text, recognizing speech,” and driving “themselves cross country.”
The history of robotics reflects this. Programmable industrial arms like Unimate (1961) succeeded early because the task was constrained and the environment was controlled. Shakey (1966-1972) showed that a mobile robot could reason about its actions, but moved at a crawl. Decades passed before consumer autonomy reached homes with the Roomba (2002) and roads with self-driving cars. Legged machines from Boston Dynamics only achieved fluid, athletic motion in the 2010s.
The throughline is that intelligence is necessary but not sufficient for a useful robot. Perception, control, mechanical reliability, and safety in an unforgiving physical world are each their own deep problem, which is why moving atoms continues to lag well behind moving bits.