The Brain: Control and Software
Sixty Years of Teaching Robots to Think
In 1961, a hydraulic arm named Unimate picked up a die-cast metal part at a General Motors plant in Trenton, New Jersey. It was the first industrial robot deployed in production, and its control system was a magnetic drum that stored a sequence of joint positions. The robot did not think. It replayed a recording. A human operator had physically guided the arm through the desired motion, the drum memorized the joint angles at each point, and the arm repeated that exact sequence thousands of times per shift.
Sixty years later, the fundamental challenge has not changed as much as you might expect. A modern industrial robot arm is faster, lighter, more precise, and vastly more reliable than Unimate. But the majority of industrial robots in operation today still run pre-programmed routines. A human specifies the motion. The robot executes it. The cleverness lives in how efficiently and reliably the robot follows instructions, not in the robot figuring out what to do on its own.
Continue reading with Explorer access
This module is part of Robotics.
Full access to all Cores, Impact Studies, Stories, and Conversations.
