Theoretical Background: History and Evolution of Robotics

Module 1 Theory: Introduction and Basics — Lecture 1

1.1 Timeline of Robotics Development

1.1.1 Ancient and Early Automata (1000 BCE - 1900 CE)

The concept of artificial beings dates back to ancient civilizations. These early efforts established the foundational idea that machines could replicate human motion and perform useful tasks autonomously.

Archytas of Tarentum (400 BCE) reportedly built a mechanical bird powered by steam, considered one of the first documented autonomous machines
Al-Jazari (1206 CE) designed programmable automata including humanoid musicians and water-powered clocks with automated figures
Leonardo da Vinci (1495) designed a mechanical knight capable of sitting, raising its visor, and moving its arms — an early humanoid robot concept
Jacques de Vaucanson (1738) created the "Digesting Duck" and a mechanical flute player, demonstrating sophisticated mechanical automation
Nikola Tesla (1898) demonstrated a radio-controlled boat, one of the first remotely operated machines

1.1.2 The Birth of Modern Robotics (1920 - 1960)

1920: Karel Capek introduced the word "robot" (from Czech robota, meaning forced labor) in his play R.U.R.
1942: Isaac Asimov published the Three Laws of Robotics — foundational ethical framework still referenced in modern HRI research
1948: William Grey Walter built Elmer and Elsie — autonomous tortoise robots demonstrating complex behavior from simple circuits
1954: George Devol filed the patent for Unimate, the first programmable robot
1956: Devol and Joseph Engelberger founded the world’s first robot company

Key Insight: Walter’s tortoises are historically significant because they demonstrated that complex emergent behavior could arise from simple reactive control — a principle that later influenced subsumption architecture and behavior-based robotics.

1.1.3 Industrial Robotics Revolution (1960s - 1980s)

1961: Unimate installed at General Motors for die-casting and spot welding — the first industrial robot in production
1969: Victor Scheinman developed the Stanford Arm, a 6-DOF electrically powered manipulator that became a standard for robotic research
1973: KUKA developed the FAMULUS, the first industrial robot with six electromechanically driven axes
1978: PUMA (Programmable Universal Machine for Assembly) developed by Unimation, becoming the workhorse of manufacturing
1979: SCARA (Selective Compliance Assembly Robot Arm) developed in Japan for high-speed assembly tasks

Defining Characteristics of This Era:

Fixed base, repetitive tasks
High precision and repeatability
Structured environments with safety cages
Pre-programmed trajectories with no sensor feedback

1.1.4 Mobile Robotics Revolution (1980s - 2000s)

1966-1972: Shakey the Robot (SRI International) — first mobile robot to reason about its own actions using AI planning
1986: Honda began development of humanoid robots, leading to ASIMO (2000)
1997: NASA’s Sojourner rover successfully explored Mars — the first mobile robot on another planet
1997: RoboCup initiative launched — using soccer as a benchmark for mobile robot research
2002: iRobot launched Roomba — bringing autonomous mobile robots into homes

Key Shift: The transition from industrial to mobile robotics required solving fundamentally different problems — sensor-based navigation, autonomous decision-making, operation in unstructured environments, and real-time obstacle avoidance.

1.1.5 Modern Collaborative and Autonomous Systems (2000s - Present)

2004: DARPA Grand Challenge accelerated autonomous vehicle research
2008: Universal Robots introduced the first commercial collaborative robot (cobot)
2012: Boston Dynamics demonstrated BigDog and later Spot — advanced dynamic locomotion
2013: Amazon acquired Kiva Systems for warehouse automation (now Amazon Robotics)
2015: ROS 2 development began, addressing real-time and industrial requirements
2020s: Growth in autonomous delivery robots, surgical robots, and AI-powered manipulation

1.2 Key Innovations and Milestones

Each decade introduced a paradigm shift in what robots could do and where they could operate:

1960s — Industrial manipulators: Automated manufacturing with fixed, repetitive tasks
1970s — Computer-controlled arms: Enabled flexible automation through programmable trajectories
1980s — Sensor integration: Introduced adaptive behavior and closed-loop control
1990s — Mobile platforms: Expanded robotics into exploration and service domains
2000s — Humanoids and cobots: Enabled human-robot collaboration in shared workspaces
2010s — Deep learning + robotics: Brought intelligent perception and manipulation
2020s — Foundation models + ROS 2: Moving toward general-purpose autonomy

1.3 Current State of the Field

1.3.1 Major Application Domains

Manufacturing: Collaborative robots working alongside humans, flexible manufacturing cells, and quality inspection using machine vision.

Healthcare: Surgical robots (da Vinci system — over 7 million procedures performed), rehabilitation robotics, and telepresence robots for remote medical consultation.

Logistics and Warehousing: Autonomous mobile robots (AMRs) for material handling, last-mile delivery robots, and automated guided vehicles (AGVs).

Agriculture: Autonomous harvesting robots, precision spraying and weeding, crop monitoring drones.

Exploration: Mars rovers (Curiosity, Perseverance), deep-sea exploration robots, search and rescue robots.

1.3.2 Enabling Technologies

Sensors: LiDAR, depth cameras (Intel RealSense), IMUs, force/torque sensors
Computing: GPU-accelerated perception, edge computing, FPGA-based control
Software: ROS/ROS 2 ecosystem, simulation tools (Gazebo, Isaac Sim)
AI/ML: Deep reinforcement learning, computer vision, natural language processing
Communication: 5G for low-latency teleoperation, cloud robotics

1.4 Future Trends and Challenges

1.4.1 Emerging Trends

Soft robotics: Compliant actuators and structures for safe interaction
Swarm robotics: Coordination of large numbers of simple robots
Bio-inspired robots: Mimicking biological systems for locomotion and manipulation
Human-robot interaction (HRI): Natural language, gesture, and intent recognition
Digital twins: Real-time simulation paired with physical robots

1.4.2 Grand Challenges

Generalization: Moving from task-specific to general-purpose robots
Safety: Ensuring safe operation in unpredictable environments
Energy: Improving battery life and energy efficiency
Ethics: Addressing job displacement, privacy, and autonomous decision-making
Interoperability: Standardizing communication and integration (ROS 2 as middleware)

1.5 Integration: Theory to Practice

The historical progression of robotics directly informs how modern systems are designed and implemented:

From Unimate to ROS 2: Early industrial robots used proprietary controllers with hardcoded trajectories. Modern systems use standardized middleware (ROS 2) that separates control logic from hardware, enabling code reuse across platforms.
From Shakey to SLAM: Shakey’s symbolic AI planner evolved into probabilistic methods. Today’s mobile robots use EKF-SLAM and particle filters — direct descendants of the sensor-based navigation pioneered in the 1970s-80s.
From safety cages to cobots: The shift from isolated industrial cells to collaborative workspaces required force-limited joints and real-time collision detection — concepts rooted in impedance control theory.
From teleoperation to autonomy: The autonomy spectrum (teleoperated, semi-autonomous, fully autonomous) maps directly to the sense-plan-act loop complexity required in the software architecture.

1.6 Theoretical Design Choices

Why ROS 2 over proprietary frameworks? ROS 2 provides a publish-subscribe middleware with DDS (Data Distribution Service) that handles real-time communication, quality-of-service policies, and multi-robot coordination. Proprietary systems lock you into vendor-specific hardware; ROS 2’s abstraction layer allows the same navigation stack to run on a TurtleBot, a warehouse AMR, or a custom platform.

Why probabilistic methods dominate modern robotics? Real sensors are noisy and environments are uncertain. Deterministic approaches (like Shakey’s symbolic planner) fail in unstructured settings. Bayesian estimation provides a principled framework for reasoning under uncertainty — this is why EKF, particle filters, and probabilistic SLAM are the standard.

Why 6-DOF for general manipulation? A task in 3D space requires specifying position (3 DOF) and orientation (3 DOF). The Stanford Arm’s 6-DOF design established this as the minimum for arbitrary spatial tasks. Redundant manipulators (7+ DOF) add flexibility for obstacle avoidance but increase computational complexity.

Why collaborative robots use force limiting rather than speed limiting alone? Impact force depends on both velocity and effective mass. ISO/TS 15066 specifies biomechanical limits on transient and quasi-static contact forces. Force-limited joints with compliant control provide safety across the full range of operating conditions, not just at reduced speeds.