Robots · Menlo Park · 1966–1972

The Robot That Needed to Think Before It Moved

The challenge was not just moving. It was deciding what movement came next.

1. A robot with a modest room

During the 1960s, researchers at SRI began building a mobile robot that could connect perception, reasoning and action. The machine operated in a controlled environment rather than an unpredictable city. Its ambition was still substantial: it would work out steps toward a goal instead of receiving every movement individually.

The challenge was not just moving. It was deciding what movement came next.

2. The name was not accidental

Shakey acquired its name from its unsteady-looking motion. Its upright body, wheels, camera and sensors made it recognisably unlike a person. The project did not depend on a convincing human disguise. It asked whether a machine could understand enough about its surroundings to accomplish useful tasks.

The robot looked shaky because looking graceful was not the main experiment.

3. Seeing becomes information

A television camera and other sensors supplied information about the room. That information had to be converted into a representation the planning system could use. Detecting an obstacle was only helpful if the robot could connect the observation with decisions about where and how to move.

A camera image was the beginning of a problem, not its solution.

4. A goal rather than a script

Researchers could give Shakey a task at a higher level than individual wheel movements. The system then needed to organise actions that would achieve it. This distinction connected the physical robot with research in artificial intelligence, where reasoning about sequences was itself a major challenge.

The instruction described an outcome. The robot had to fill in the journey.

5. Planning changes the route

If completing a task required moving around an obstacle or interacting with an object, the actions had to fit together. A reasonable first move could still leave the robot unable to finish. Planning meant considering the relationship between immediate steps and the intended final state.

Going somewhere was easier than knowing why that was the right place to go.

6. The room helped the experiment

Shakey's environment was simplified and carefully arranged for research. That limitation was not a secret demonstration of human-level understanding. It made the problem manageable enough to investigate. Explaining the controlled conditions gives the achievement its proper scale rather than shrinking or exaggerating it.

The room was simple so the underlying question could be difficult.

7. Computing did not fit inside

Much of the processing depended on computers outside the mobile robot, connected through communications equipment. The visible machine was therefore only part of the system. Its importance lay in how those components worked together, not in pretending that all the intelligence lived inside a single rolling box.

The robot's apparent body was smaller than its actual system.

8. Software survives the hardware

Work associated with Shakey contributed to important approaches in search and planning, including systems that influenced later artificial-intelligence research. The lasting achievement was not limited to one machine's movements. Ideas developed while solving its tasks could be reused in very different settings.

The robot's most influential descendants were sometimes algorithms.

9. A research project becomes history

The principal Shakey project ran from 1966 to 1972. The machine later became a museum object and a landmark in robotics history. Displaying the hardware can make the project look quaint, but its combination of perception, planning and action was a consequential step.

The old box on wheels had asked a surprisingly modern question.

10. The useful kind of ambition

Shakey did not demonstrate a machine capable of navigating all human life. It showed that carefully bounded physical tasks could connect with symbolic reasoning. Remembering both the breakthrough and the boundaries explains why the project mattered without turning a research robot into a fictional all-purpose intelligence.

It did not need to think like a person to change how robots were built.