Other Machines — Day 3

Shakey Robot

XINDI YU
2 min readFeb 12, 2018

Meet “Shakey” — the world’s first robot to embody artificial intelligence. Shakey could perceive its surroundings, logically deduce implicit facts from explicit ones, navigate from place to place, make a plan to achieve a goal, monitor the execution of a plan in the real world, recover from errors in plan execution, improve its planning abilities through learning, and communicate in simple English.

Software and Hardware

Shakey, which was named after its wobbly gait, had a TV camera, a triangulating range finder, and bump sensors, and was connected to DEC PDP-10 and PDP-15 computers via radio and video links. Shakey used programs for perception, world modeling, and acting. Low-level action routines took care of simple moving, turning, and route planning. Intermediate level actions strung the low-level ones together in ways that robustly accomplished more complex tasks. The highest-level programs could make and execute plans to achieve goals given it by a user. The system also generalized and saved these plans for possible future use.

Originally, Shakey was controlled by a SDS-940 computer acquired in 1966 with 64K 24-bit words of memory. Programmed in Fortran and Lisp, Shakey’s problem solving was done in QA3. This was replaced by a “large” PDP-10 around 1969 with 192K 36-bit words of memory. STRIPS was then used for problem solving, with QA4 developed later. When the movie was made, Shakey’s programs occupied over “300,000 36-bit words” (~1.35 MB).

Images showing SHAKEY’s visual perception capabilities using a region-analysis method developed by SHAKEY design team members Claude Brice and Claude Fennema: (1) an array of pixels from a digitized TV image, (2) the pixels grouped into regions, (3) similar regions merged, and (4) lines fit to resulting boundaries and shapes, which are thereby recognized as a cube, wedge, floor, wall, etc. Images and descriptions from SHAKEY design team members Claude Fennema and Richard Duda.

--

--