Duration: 120 mins BCSE427L Cognitive Robotics Dr. V.
Arunkumar
Module 1: Introduction
1. Cangelosi and M. Asada, Cognitive Robotics Handbook, MIT Press, 2022; Chapter 1
2. G. Sandini, A. Sciutti, and D. Vernon. Cognitive Robotics. In M. Ang, O. Khatib, and B.
Siciliano (Eds.), Encyclopedia of Robotics, Springer, 2021.
Session Overview
• Introduction
• Thinking
• Cognition, and Intelligence
• Defining Intelligence – Embodiment and Its Implications
• Synthetic Methodology for Intelligence
Introduction
What is Cognitive Robotics?
"Cognitive Robotics is the field that combines
insights and methods from robotics, AI and
cognitive and biological sciences to design an
integrated cognitive system combining the
sensorimotor behavior and higher-level functions
and social capabilities of an intelligent robot"
…Continued
Evolution
What’s Cognition?
“Cognition is the process by which an
autonomous system perceives its
environment, learns from experience,
anticipates the outcome of events, acts
to pursue goals, and adapts to changing
circumstances.”
Cognition and Thinking
• Cognition is thinking
• But it also encompasses the processes associated with perception, knowledge, problem
solving, judgment, language, and memory
• Thinking is an important component, but cognition also encompasses unconscious and
perceptual processes as well.
• In addition to thinking, cognition involves language, attention, learning, memory, and
perception
Cognition and Intelligence
• Cognition is a broader term that encompasses the various mental processes that
contribute to intelligence
• While intelligence is often associated with the ability to perform well on
standardized tests or demonstrate exceptional academic or analytical skills,
cognition refers to the underlying mechanisms and mental processes that
enable these abilities
• Cognition is the foundation upon which intelligence is built, as it involves the
fundamental mental operations that allow an individual to acquire, manipulate,
and utilize knowledge.
• Intelligence is the overall capacity to think, learn, and solve problems, while
cognition is the specific set of mental processes and abilities that enable these
higher-order functions
Why is cognition useful in robotics?
• It allows robots to work autonomously in everyday environments
• It enables effective interaction with humans
Modelling Cognition
• First step in developing artificial cognitive systems
1. Purpose of Cognition
2. Mechanisms by which cognitive systems fulfil the purpose
Aspects of Modelling Cognitive Systems….
1. Computational Systems
2. Bio-Inspired Systems
3. Hybrid
Computational / Biological
Level of Abstraction
Mutual Dependence of Brain, Body and Environment
• Embodied Cognition
• In human cognition, Example of embodied cognition
is to Gesture of hands while talking
Ultimate – Proximate Distinction
• “Why Vs How “
• Ultimate explanations deal with questions concerned with why a given behaviour exists in a
system or is selected through evolution
• While proximate explanations address the specific mechanisms by which these behaviours
are realized
Cognition Cycle
More Refined Definition of ”Cognition”
• Cognition is the process by which an autonomous system perceives its environment, learns
from experience, anticipates the outcome of events, acts to pursue goals, and adapts to
changing circumstances
Why Autonomy?
• To increase the system’s repertoire of effective actions, and
• To extend the time-horizon of its ability to anticipate the need for and outcome of future
actions
Desirable Features of Cognitive Systems
• Realism
• Behaviour – not exceedingly complex
• Functional capabilities
• Dynamics
• Development
Embodied Cognition
What is it?
Many features of cognition are embodied in that they are deeply dependent upon characteristics
of the physical body of an agent, such that the agent’s beyond-the-brain body plays a significant
causal role, or physically constitutive role, in that agent’s cognitive processing
How it affects the system?
• Through the physiology of the cognitive system
• Through its evolutionary history
• Through practical activity
• Through its socio-cultural situation
Through Physiology
• An agent’s perceptions depend not just on what’s happening in the environment and what the
senses convey about these events, but they also depend on the state of the motor circuits
• This two-way dependency between the sensory states and motor states extends further to include
also a two-way dependency between perceptions (and the concepts they convey) and actions (and
the goals implied by these actions)
• The mutual dependence of perception and action implies a dependence of cognition on the
embodiment of the cognitive agent and the actions that embodiment enables.
Through Evolutionary History
• The way we infer things can be traced back to more primitive inference mecha- nisms that derive
from an earlier phase in the history of evolution
• This is like redeploying an ultimate capability for a different new purpose
Through practical activity
• As a robot engage in practical activity, it develops an understanding of the environment in terms of
your embodied action capabilities
Socio-Cultural situatedness
• With respect to robots, it is the previous learning it had through its interactions with other entities
Three Hypotheses on Embodiment
1. Conceptualization Hypothesis
• The position that the physical morphology — the shape or form — and motor capabilities of a
system has a direct bearing on the way the cognitive agent understands the world in which it is
situated
2. Context Hypothesis
• The idea that the body (and possibly also the environment) plays a constitutive rather than a
supportive role in cognitive processing, i.e. that the body is itself an integral part of cognition
3. Replacement Hypothesis
• Cognition is not only influenced and biased by the characteristics and states of the agent’s
body, but the body and its dynamics augment the brain as an additional cognitive resources
Minimal Embodiment
• Minimal embodiment requires that the cognitive system be capable of “full” interaction, i.e.
interaction that not only influences and alters the state of the world but, in turn,
that these influences on the environment in turn influence the interactive process of the cognitive
system
Types of Embodiment
1. Structural coupling between agent and environment
• A system can be perturbed by its environment and can in turn perturb its environment
2. Historical embodiment
• as a result of a history of structural coupling
3. Physical embodiment
• In a structure that is capable of forcible action (this excludes software agents)
4. Organismoid embodiment
• organism-like bodily form (e.g. humanoid robots)
5. Organismic embodiment
• Autopoietic living systems
6. Social embodiment
• Reflecting the role of the agent’s body in social interaction
….Continued
Methodologies to achieve
cognition in robotics
Cognitive Architecture
• The requirements of the architecture needs to be identified, partly through an analysis of
several typical scenarios in which the eventual agent would demonstrate its competence
• These requirements are then used to create an architecture schema: a task- and
implementation-independent set of rules for structuring processing components and
information, and controlling information flow
• Logical division of task-independent processing mechanisms and structure from task-
dependent knowledge
Examples of Cognitive Architecture
• Soar
• Darwin
• ISAC
Soar
• It was created by Allen Newell
• Soar is a production system
(sometimes called a rule- based
system)
• Rules in Soar are called associations
• Two memories – Recognition and
Declarative
• New production rules - chunking
Darwin
• Darwin is a series of robot
platforms designed to experiment
with developmental agents
• BBD – Brain Based Devices
• Neuromimetic: mimic the neural
structure of the brain
ISAC
• Intelligent Soft Arm Control
• A hybrid cognitive architecture for
an upper torso humanoid robot
• SAC is constructed from an
integrated collection of software
agents and associated memories
• STM has a robot-centred spatio-
temporal memory
Autonomy
• Types of Autonomy
• Energy
• Biological
• Behavioural and constitutive
• Homeostasis and Allostasis
Types of Memory
• short-term vs. long-term
• declarative vs. procedural
• semantic vs. episodic
• symbolic vs. sub-symbolic
• modal vs. amodal modal
• hetero-associative vs. auto-associative