Robotics and Autonomous Systems - Introduction

Intelligent Robots

• Definition: A physically situated intelligent agent.
• Agent:
  • Anything that can sense its surroundings and take actions.
  • A person or thing that does an action (dictionary definition).
  • Can be software, hardware (robots), or people.
• Intelligent Agent:
  • A system that perceives its environment and takes actions to maximize the chances of success.
• Robot:
  • A special type of agent that is physically situated in the real world.
  • Distinction from software agents situated in a virtual world.
  • Simulated robot: A software agent resembling an actual robot with the same control system.
• Physically Situated:
  • Being present in the physical world.
  • Distinct from situation awareness.
• Situation Awareness:
  • Understanding how the world works and predicting outcomes of actions.

Components of an Intelligent Robot

• List is arbitrary and may omit elements like mechanics (chassis).

• Includes hardware and software.

• Effectors:

  • Arms, grippers (simple robotic hands).
  • Wheels, legs, blades (e.g., quadcopter blades).
  • Hardware.

• Perception:

  • Hardware: Cameras, lidars, touch sensors.
  • Software: Algorithms processing sensor data.

• Control:

  • Hardware: Computing hardware (computer).
  • Software: Control algorithms implementing decision-making.
  • Key to a robot's intelligence; focus of the course.

• Communications:

  • Hardware: Network interface cards.
  • Software: Algorithms for communication with people, robots, and other agents.

• Power:

  • Hardware: Batteries.
  • Software: Low-level battery control to ensure continuous operation.

Types of Robots

• Classified by environment and application.

• Environment:

  • Ground.
  • Aerial.
  • Water (on or inside).
  • Classification can be ambiguous (e.g., legged robots with drones).
  • Piping inspection robots: Ground vehicles moving around water.

Biomimetic Robots and Androids

• Biomimetic Robot:
  • Inspired by or resembling animals.
• Android:
  • A biomimetic robot resembling a human.

Types of Ground, Aerial, and Water Robots

• Ground Robots:
  • Wheeled robots (including androids).
  • Legged robots (including androids).
  • Tracked robots.
  • Snake-like robots.
• Aerial Robots:
  • Fixed-wing.
  • Rotor-based (quadcopters, helicopters).
  • Wind-flapping (less common).
• Water Robots:
  • Unmanned Surface Vessels (USVs).
  • Autonomous Underwater Vehicles (AUVs).
  • AUVs often tethered to USVs or boats due to limited autonomy.

Applications of Robots

• Classified by use rather than environment.

• Tools to Replace Humans:

  • Telepresence, assistance, amusements, companionship, education, and research.

• The Three Ds:

  • Reasons to replace humans with robots: Dirty, dull, and dangerous tasks.

• Examples:

  • Pipe cleaning and inspection (dirty).
  • Vacuuming (dirty and dull).
  • Industrial robots (dull).
  • Explosive disposal/demining (dangerous).
  • Nuclear plants (dangerous).
  • Rescue robots (dangerous and dirty).

Telepresence Robots

• Personal use: Video conferencing.
• Professional use: Surgery performed remotely, military, space exploration (Moon and Mars).
• Types:
  • Zero autonomy: Manually controlled robots.
  • Taskable agents: Robots with higher autonomy given complex tasks or missions.
• Taskable Agent Definition:
  • Agent given a complex task or mission that it executes without supervision, returning after completion.
• Telepresence Challenges:
  • Improve user feedback (haptic feedback).
  • Reduce communication latency.
  • Reduce cognitive fatigue (caused by the difference in how robots and humans operate).

Assistance Robots

• Applications: Elder care, rehabilitation, and nursing.
• Examples:
  • Bear-like robot for children with cerebral palsy (suggesting exercises).
  • Robot delivering noodles based on voice command and environmental perception.
• Challenges:
  • Perception.
  • Context awareness.
  • Human-robot interaction.
  • Automated task planning.

Amusement and Companionship Robots

• Therapeutic robots: Paro (basic pet-like behaviors, seeking affection).
• Companion robots: AIBO, Vector (basic behaviors, embodied Alexa-like features). Often have limited lifespans due to business decisions.
• Multipurpose companionship: Revit (mobile, automatic system for video conferencing, object location, user monitoring, and games).

Robotics in Education and Research

• Education:
  • Teaching programming, electronics, mechanics, mathematics, trigonometry.
  • Commercial robots: Stimio, Cosmo.
• Research:
  • Applicable to any area of research in AI and robotics.

Seven Areas of AI

• Arbitrary areas, can be merged or expanded.

• Knowledge Representation:

  • Representing the world, context, goals of agents (including humans), and the robot's own state.

• Understanding Natural Language:

  • Understanding meaning of words in context, filtering background noise, locating voices.

• Learning:

  • About the environment (mapping, dynamics).
  • Identifying users.
  • Imitating humans.
  • Adapting to user preferences.

• Planning/Problem Solving:

  • Achieving goals with plans.
  • Essential for all robotic applications except those that are completely teleoperated.

• Behaviourism:

  • Belief that mental processes can only be studied through actions (physical actions, fMRI).
  • $(fMRI)$ is functional magnetic resonance imaging.
  • Introspection seen as of little use, or no use, to analyze or understand intelligence.

• Inference:

  • Generating an answer when there is no complete explicit information.
  • Using information when the data isn't directly provided.
  • Moving from premises to logical consequences.
  • Related to planning; perceptual plans can involve inference.
  • Inference is useful when you
    child(a) \and child(c) \and sibling(a,b) \implies sibling(b,c)
    

• Search:

  • Search in a search space (e.g., chess moves in a state space).
  • Searching solutions, or near optimal actions.

• Vision:

  • Achieving understanding from images.
  • Object detection and recognition.
  • Scene recognition.
  • Action recognition.
  • Identification of human users.

History of Robotics

• Evolution from tools to agents/companions.
• Interrelation of AI and robotics.
• Early Robots:
  • Teleoperated robots (Waldos) for manipulating radioactive materials.
  • Drones for target practice.
• Industrial manipulators and mobile robots:
  • Industrial robots largely preprogrammed lacking a lot of autonomy. Mobile robots had slightly higher autonomy.
• Industrial Manipulators:
  • Reprogrammable, multifunctional mechanisms for moving materials or tools.
  • High number of degrees of freedom necessary.
  • Precise, durable, low maintenance, relatively low power consumption.
• Degrees of Freedom (DOF):
  • The number of independent joint positions needed to specify the exact configuration of a robot.
  • $DOF$ is the short form
  • Challenges includes that industrial manipulators must be profitable.
• Redundancy in Manipulation:
  • The phenomena that happens when the position of the effector can be achieved with different possible configurations.
• Control Systems:
  • Open loop: Commands sent without feedback.
  • Closed loop: Continuously comparing actual and desired configuration, minimizing the difference.
• Drones
  • Originally remote controlled planes for target acquisition.
  • Evolved to carry weapons.
  • Now used for land monitoring and fruit picking.
  • Still an umbrella term for unmanned aerial robots.
• Mobile Robots
  • They must have the capability to clean industrial complexes and have been used for transport.

State of Art

• Mobile robots integrate into several real world scenarios and integrate together with human co-workers.