Home · Module 1 · History & Evolution
History & Evolution of Robotics
From mythological automata to autonomous humanoids — click any era to expand. The same engineering questions keep coming back: how to perceive, decide, and move.
Core idea — Robotics is the convergence of three traditions: mechanical engineering (motion), control theory (stability), and computer science (decision-making). Each era pushed one tradition further.
Ancient automata
Hero of Alexandria's steam-powered devices and self-opening temple doors.
The earliest documented "robots" were pneumatic and hydraulic automata — programmable by physical pegs and cams, not code. Hero's Pneumatica describes wind-powered organs and coin-operated holy water dispensers. These machines established the principle of stored programs two millennia before Turing.
PRINCIPLE Sequential mechanical control via cams = the great-great-grandfather of the modern state machine.
Leonardo's mechanical knight
Da Vinci's humanoid robot — gears, pulleys, cables.
Designed for a Milan court pageant, Leonardo's knight could sit, stand, raise its visor, and move its arms. Driven by a system of pulleys, cables, and a programmable cylinder cam — essentially the first humanoid kinematic chain.
RELATES TO Module 1 — Kinematics (DOF, linkages, joint coordination).
The word "robot" is born
Karel Čapek's play R.U.R. coins robota (forced labor).
The Czech term robota means "drudgery" or "forced labor." Čapek's play imagined synthetic humans manufactured for cheap labor who eventually rebel — establishing both the cultural archetype and the ethical anxieties we still grapple with today.
RELATES TO Course topic: Ethical Considerations of Robotics & AI in Society.
Asimov's Three Laws
The first formal ethics framework for autonomous machines.
In Runaround, Isaac Asimov published the Three Laws of Robotics:
- A robot may not injure a human being or, through inaction, allow a human to come to harm.
- A robot must obey orders given by humans except where such orders conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Real safety-critical robotics still uses descendants of this — safety hierarchies, watchdog timers, and emergency-stop overrides.
Cybernetics & Grey Walter's tortoises
Norbert Wiener formalizes feedback. Walter builds Elmer and Elsie.
Wiener's Cybernetics defined the mathematics of feedback loops in animals and machines — the theoretical bedrock of PID control. The same year, W. Grey Walter built two tortoise-like robots that exhibited goal-seeking, light-avoiding, obstacle-avoiding behavior — proving that simple sensorimotor loops can generate apparently intelligent behavior.
RELATES TO Module 2 — Control Systems.
Unimate — the first industrial robot
George Devol's hydraulic arm joins a General Motors assembly line.
Unimate weighed 1,800 kg and could handle red-hot die-cast parts that were dangerous for humans. It launched the modern industrial robot industry — by 2024, over 4 million such robots are operating worldwide.
PRINCIPLE Repeatability + accuracy + 24/7 operation = the economic case for robots that hasn't changed in 60 years.
Shakey at Stanford
The first general-purpose mobile robot with vision and reasoning.
Shakey integrated perception (TV camera), planning (STRIPS — a foundational AI planning algorithm), and action (motorized base) in a single system. It also pioneered the A* search algorithm, still used in pathfinding today (yes, including in your Raspberry Pi car).
RELATES TO Module 1 — Robotics & AI. Module 4 — Autonomous Navigation.
Subsumption architecture
Rodney Brooks: "Intelligence without representation."
Brooks rejected the heavy "plan-then-act" paradigm and showed that layered reactive behaviors could produce robust mobility. His insectoid robots could navigate cluttered offices without any internal world model. This philosophy underpins modern behavior-based robotics.
RELATES TO Module 4 — Reactive obstacle avoidance.
Honda ASIMO
Stable, dynamic bipedal walking goes mainstream.
ASIMO ("Advanced Step in Innovative Mobility") demonstrated reliable dynamic bipedal locomotion — running, climbing stairs, recognizing faces. Its Zero-Moment Point (ZMP) control technique is still standard for humanoid balance.
SoftBank Pepper
The first mass-market humanoid designed to read emotion.
Pepper combines a 3D camera, microphones, touch sensors, and a tablet to interact with people in retail, banking, and education. Programmed via Choregraphe (visual block-based) or NAOqi (Python/C++). Pepper is the humanoid you will program in this course.
RELATES TO Module 3 — Pepper Robot, NLP for Robots.
Deep learning takes over
End-to-end neural policies, sim-to-real transfer, foundation models.
AlphaGo (2016), then OpenAI's robotic Rubik's Cube hand (2019), then Boston Dynamics' parkour-running Atlas (2021), then RT-2 (2023) — a foundation model that maps vision + language directly to robot actions. Robotics has moved from hand-engineered control to learned policies.
PRINCIPLE Today: every robot is a hybrid — classical control for stability, learned models for perception and high-level decisions.
You are here
Your Raspberry Pi car competition is the latest entry in this lineage.
The autonomous obstacle-avoiding car you will build in Module 4 packages 2,500 years of robotics history into a $50 platform: sensors (ultrasonic — Hero would recognize the principle), actuators (DC motors — Devol's territory), control (PWM, feedback — Wiener's mathematics), perception & decision (your Python code — Shakey's heritage).
That's the point of the course: you don't just use robotics, you are the next step in its history.