Unit 6: Intelligent Robotics and Applications

I. Orientation — Foundations of Intelligent Robotics

Intelligent robotics combines mechanical systems, sensors, computation, communication, and artificial intelligence (AI) to create machines that can perceive conditions, select actions, and operate with varying degrees of autonomy.

• Governing principle: A robot repeatedly executes a sense–plan–act cycle: sensors gather data, algorithms interpret it, and actuators change the robot or environment.
• Core components:
  • Sensors: Cameras, encoders, force sensors, LiDAR, microphones, and inertial measurement units.
  • Controller: A microcontroller, programmable logic controller (PLC), or computer running control and AI algorithms.
  • Actuators: Electric motors, pneumatic cylinders, hydraulic drives, or artificial muscles.
  • End effectors: Grippers, welding torches, surgical tools, sprayers, or other task-specific devices.
• Feedback principle: Closed-loop control compares desired output (r(t)) with measured output (y(t)), producing error (e(t)):

e(t) = r(t) - y(t)

• Levels of operation: Robots range from fixed, preprogrammed industrial machines to autonomous systems that adapt their actions from sensor data.
• Performance criteria: Accuracy, repeatability, response time, payload, energy use, reliability, safety, and task-completion rate determine suitability.

II. Machine Vision — Visual Perception for Robots

A. Machine vision overview

Machine vision enables a robot to acquire and interpret images for inspection, recognition, measurement, localization, and control.

• Imaging chain: A typical system contains illumination, lens, camera sensor, image-acquisition hardware, processing software, and robot interface; controlled LED lighting often improves consistency.
• Processing pipeline:
  • Acquisition: A camera produces a grayscale, RGB, depth, or thermal image.
  • Preprocessing: Filtering reduces noise; calibration corrects lens distortion.
  • Segmentation: Thresholding or learned models separate objects from the background.
  • Feature extraction: Edges, corners, contours, colour, texture, or neural-network features describe objects.
  • Decision: The system classifies an item or estimates its position and orientation.
• Coordinate conversion: Camera calibration transforms image coordinates ((u,v)) into robot-frame coordinates ((x,y,z)), allowing a manipulator to grasp a detected object.
• Vision types: Two-dimensional vision supports barcode reading and surface inspection; stereo cameras, structured light, and time-of-flight sensors provide three-dimensional depth.
• Limitations: Shadows, reflections, occlusion, motion blur, poor calibration, and unfamiliar objects can reduce accuracy.

III. Artificial Intelligence — Reasoning, Learning, and Adaptation

A. Artificial intelligence in robotics

Artificial intelligence gives robots methods for perception, decision-making, learning, prediction, language interaction, and adaptation beyond rigid programming.

• AI approaches: Rule-based systems use explicit conditions, while machine learning derives patterns from training data; deep learning commonly handles images, speech, and complex sensor inputs.
• Planning: Algorithms such as A* search select a low-cost route through a map by evaluating:

f(n) = g(n) + h(n)

• (g(n)) is the known cost from the start to node (n).
• (h(n)) is the estimated cost from (n) to the goal.
  • Learning methods: Supervised learning uses labelled examples; unsupervised learning identifies structure; reinforcement learning improves a policy through rewards and penalties.
  • Robotic uses: AI supports object recognition, grasp selection, predictive maintenance, voice commands, anomaly detection, and adaptive motion.
  • Constraints: Performance depends on representative data, computational capacity, explainability, real-time response, and resistance to unexpected inputs.

IV. Autonomous Systems — Independent Perception and Action

A. Autonomous robots

Autonomous robots perform tasks with limited direct human control by sensing their environment, estimating their state, planning, and acting.

• Autonomy loop: Perception builds an environmental model; localization estimates robot pose; planning selects a route; control generates motor commands; feedback detects deviations.
• Localization and mapping: Simultaneous localization and mapping (SLAM) allows a mobile robot to construct a map while estimating its own position within it.
• Navigation: Global planning finds a route across a known map, whereas local planning avoids newly detected obstacles using LiDAR, sonar, depth cameras, or bump sensors.
• Levels of autonomy: Systems may be remotely operated, supervised, conditionally autonomous, or highly autonomous; human override remains important in safety-critical work.
• Examples: Warehouse autonomous mobile robots transport shelves, robotic vacuum cleaners map rooms, and planetary rovers execute delayed commands where continuous teleoperation is impractical.
• Challenges: Dynamic environments, sensor failure, uncertain terrain, battery limits, and unusual edge cases can interrupt autonomous operation.

V. Communication Systems — Data Exchange and Coordination

A. Robot communication

Robot communication transfers commands, measurements, status information, and coordination messages among sensors, controllers, robots, and human operators.

• Internal communication: CAN bus, EtherCAT, I²C, SPI, and serial links connect encoders, motor drives, controllers, and embedded sensors.
• External communication: Ethernet and industrial field networks provide reliable factory links; Wi-Fi, Bluetooth, cellular, and radio support mobile or remote robots.
• Communication models: Client–server exchanges direct requests, while publish–subscribe systems distribute messages by topic; the Robot Operating System (ROS) commonly uses nodes, topics, services, and actions.
• Performance measures: Latency is transmission delay, bandwidth is data capacity, jitter is delay variation, and packet loss is the fraction of data that fails to arrive.
• Multi-robot coordination: Robots share position, task, and obstacle data to prevent collisions and allocate work.
• Security requirements: Authentication, encryption, network segmentation, access control, and signed updates reduce spoofing, interception, and unauthorized control.

VI. Connected Robotics — Cloud and Internet Integration

A. Internet of Robotic Things (IoRT)

The Internet of Robotic Things combines connected robots, IoT sensors, cloud or edge computing, and intelligent services for distributed monitoring and control.

• Architecture: Physical robots generate data; gateways connect devices; edge computers handle time-critical processing; cloud platforms store data and perform large-scale analytics.
• Capabilities: IoRT enables remote diagnostics, fleet management, software updates, digital twins, usage tracking, and predictive maintenance based on vibration, temperature, or motor-current trends.
• Edge–cloud contrast:
  1. Edge computing: Processes data near the robot, reducing latency and dependence on continuous connectivity.
  2. Cloud computing: Provides extensive storage and computation but introduces network delay and availability concerns.
• Example: A warehouse platform can assign delivery tasks to multiple robots while tracking battery charge and congestion through a central fleet server.
• Risks: Expanded connectivity increases exposure to cyberattacks, privacy breaches, incompatible protocols, communication failure, and unsafe remote commands.

VII. Human–Robot Collaboration — Shared Workspaces

A. Collaborative robots (Cobots)

Collaborative robots are designed to work near or with people under defined risk controls rather than remaining permanently isolated behind fences.

• Collaborative features: Force–torque sensing, rounded surfaces, lightweight links, speed monitoring, compliant joints, and emergency stopping reduce contact hazards.
• Operating modes: Safety-rated monitored stop, hand guiding, speed-and-separation monitoring, and power-and-force limiting support different forms of collaboration.
• Programming: Operators may guide a cobot arm through desired poses or use graphical task blocks, reducing programming effort for small-batch production.
• Applications: Cobots perform machine tending, screwdriving, packaging, inspection, laboratory handling, and repetitive assembly while humans manage judgement-intensive steps.
• Advantages: Flexible deployment, smaller floor-space requirements, and easier task changeover suit low-volume, high-mix manufacturing.
• Limitations: Safety restrictions generally reduce speed and payload; sharp tools, heavy workpieces, and pinch points still require guarding and formal risk assessment.

VIII. Safe Robotic Operation — Hazard Control

A. Safety in robotics

Robotic safety aims to prevent injury, equipment damage, and environmental harm throughout design, installation, operation, maintenance, and disposal.

• Major hazards: Crushing, trapping, impact, unexpected startup, electrical shock, hydraulic release, dropped payloads, software faults, and loss of communication.
• Risk process: Identify hazards, estimate severity and probability, apply controls, verify performance, and document residual risk.
• Control hierarchy: Eliminate hazards first, then use engineering controls such as guards and interlocks, administrative procedures, training, and personal protective equipment.
• Protective devices: Emergency stops, light curtains, laser scanners, safety mats, limit switches, brakes, and safe-torque-off circuits can halt dangerous motion.
• Safety standards: ISO 10218 addresses industrial robot safety, while ISO/TS 15066 provides guidance for collaborative robot applications.
• Operational discipline: Lockout/tagout isolates energy during maintenance; safety functions must be tested after programming, tooling, or layout changes.

IX. Responsible Robotics — Social and Moral Constraints

A. Ethics in robotics

Ethics in robotics examines how robotic systems affect human rights, responsibility, fairness, privacy, employment, and the use of force.

• Human accountability: Designers, manufacturers, deployers, and operators must retain responsibility; describing an AI system as autonomous does not remove human obligations.
• Bias and fairness: Vision or decision systems trained on unrepresentative data may perform unevenly across environments or populations, requiring dataset evaluation and outcome monitoring.
• Privacy: Domestic, medical, and workplace robots may collect video, audio, location, and health information; data minimization and informed consent reduce misuse.
• Transparency: Users should know when they interact with a robot, what information it collects, and the basis and limits of safety-critical decisions.
• Employment effects: Automation can remove hazardous or repetitive work but may displace roles, alter skills, and require retraining.
• Dual use: Navigation, surveillance, and manipulation technologies may support rescue or weapon systems, making proportionality, meaningful human control, and legal review essential.

X. Manufacturing Systems — Automated Production

A. Industrial automation

Industrial automation uses robots and control systems to perform production operations with consistent speed, precision, and repeatability.

• Automation hierarchy: Sensors and actuators form the field level; PLCs control machines; supervisory control and data acquisition monitors processes; manufacturing execution systems coordinate production.
• Robot applications: Spot welding, arc welding, painting, palletizing, pick-and-place, assembly, dispensing, and machine loading are common operations.
• Robot selection: Payload, reach, number of axes, cycle time, repeatability, environmental rating, and end-effector compatibility determine system design.
• Integration: A PLC may command the robot, conveyor, vision system, and safety controller through an interlocked sequence.
• Benefits: Automation increases throughput, product uniformity, traceability, and worker separation from heat, fumes, chemicals, or heavy loads.
• Limitations: High initial investment, specialist maintenance, production downtime, and limited flexibility can make automation unsuitable for unstable or very low-volume processes.

XI. Sector-Specific Applications — Robotics Beyond Factories

A. Medical, agricultural, logistics, and defense robotics

Sector-specific robots adapt sensing, mobility, tools, and autonomy to environments with different technical, legal, and safety demands.

• Medical robotics: Surgical systems translate a clinician’s commands into precise instrument motion; rehabilitation exoskeletons assist movement, while service robots transport medicines and samples. Sterility, reliability, and clinical validation are critical.
• Agricultural robotics: GPS, cameras, multispectral sensors, and machine learning support autonomous tractors, precision spraying, weeding, harvesting, and crop monitoring; irregular terrain and changing weather remain challenges.
• Logistics robotics: Automated guided vehicles follow predefined routes, whereas autonomous mobile robots dynamically navigate warehouses; robotic arms sort parcels using barcode readers or vision.
• Defense robotics: Uncrewed ground, aerial, surface, and underwater vehicles perform reconnaissance, bomb disposal, logistics, surveillance, and search in hazardous areas.
• Comparative requirement: Medical robots emphasize patient safety, agricultural robots need outdoor robustness, logistics robots prioritize throughput, and defense systems require secure communication and operation under uncertainty.
• Human oversight: High-consequence decisions, especially clinical actions and use of force, require clear authority, traceability, and appropriate human control.

XII. Future Directions — Developing Robotic Capabilities

A. Emerging trends in robotics

Emerging robotics research is producing more adaptable, connected, energy-efficient, and human-aware systems.

• Embodied AI: Robots increasingly learn through physical interaction, combining vision, language, touch, and action rather than processing isolated data.
• Foundation models: Large multimodal models can interpret natural-language instructions and generalize across tasks, though grounding errors and unpredictable outputs require safeguards.
• Soft robotics: Flexible pneumatic, elastomeric, or tendon-driven structures enable safer handling of delicate objects and adaptation to irregular shapes.
• Swarm robotics: Many simple robots coordinate through local rules, offering scalability and fault tolerance for exploration, monitoring, or collective transport.
• Digital twins: Virtual models mirror physical robots for simulation, commissioning, maintenance prediction, and optimization before changes reach real equipment.
• Advanced sensing and mobility: Tactile skins, event cameras, improved prosthetics, legged locomotion, and bio-inspired designs extend operation in unstructured environments.
• Sustainable robotics: Energy-efficient actuators, recyclable materials, repairable designs, and battery management reduce environmental cost across the robot life cycle.