IWKS 2300 Fall 2019 Artificial Intelligence and Robotics · 2019. 11. 9. · • 90’s:Rational...
Transcript of IWKS 2300 Fall 2019 Artificial Intelligence and Robotics · 2019. 11. 9. · • 90’s:Rational...
IWKS 2300
Fall 2019
Artificial Intelligence and
Robotics
John K. Bennett
First, a little clean up from last class…
What we need to know:
1.Vfwd of LED
2.High (‘1’) voltage level of Ardunio “high”
3.Max and working high current of Arduino output
Let’s take these each in turn:
R What is the correct value of R?
A little clean up from last class…What we need to know:
1.Vfwd of LED
Vfwd (white LED at 10 ma) = 3.2V
A little clean up from last class…What we need to know:
2.High (‘1’) voltage level of Ardunio “high
V’1’ (10 ma at 25 deg. C) = 4.75V
A little clean up from last class…What we need to know:
3.Max and working high current of Arduino output
Imax = 40 ma
VoutH (5V VCC) = 4.2V
Split the difference and
make it 4.5V
A little clean up from last class…Now compute R:
R = V / I = (4.5 – 3.2) / .010ma
= 1.3V / .010
= 130 ohms (this is a min value; other color LEDs
have much lower Vfwd, e.g. red LED has a Vfwd of
1.3V: (4.5 – 1.3) / .010 = 320 ohm.
A white LED with 330 ohm R has a Ifwd of 4ma
So 270 ohm or 330 ohm gives us a safe
cushion for any LED, e.g.:
Vfwd (IR LED at 10 ma) = .9V
R = (4.5 – 0.9) / .010 = 360 ohm;
with 330 ohm R we get 11 ma (no problem)
with 130 ohm R we get 27 ma (not good!)
Now back to our regularly
scheduled programming…
What is Intelligence?
• Examples of intelligent
processes:
• Deduction
• Induction and learning
• Making rational decisions
• Intelligent embodied agents can:
• Perceive the world around them
• Act in the world
Examples of “Natural”
Intelligence
Viruses
• not considered by most biologists to be alive
• do not reproduce on their own, but make
bacteria produce copies of themselves
• no learning in individual viruses; only
“evolutionary learning”
Bacteria
• are considered by most biologists to be alive
• reproduce by dividing into two organisms
• capable of some simple learning that allows
them to move toward favorable environments
Insects
• capable of learning and memorizing
• primitive social interactions (trail of pheromones)
• simple visual and olfactory perception (compound eyes)
• coordinated movements of legs to master different types of terrains
Reptiles
• excellent vision and eye movement
• able to learn certain behaviors
• excellent hunting abilities
• very primitive social interactions
Higher Mammals
• powerful senses
• produce sounds but not language
• complex social interactions
• probably have consciousness, and in
possession of basic feelings that are related
to our own
Non-Human Primates
• human-like genetic makeup and behaviors
• produce different sounds with different meanings
• capable of learning symbolic “language”
• probably have consciousness, are self-aware and
possess basic feelings at the level of a human child
Humans
• complex social behaviors
• able to learn high-level syntactic languages
• extremely long phase of learning (upbringing)
• consciousness, self-awareness, abstract thinking
• awareness of past and future, planning capability
Hardware
1011 neurons
1014 synapses
cycle time: 10-3 sec
Hardware
1011 neurons
1014 synapses
cycle time: 10-3 sec
1010 transistors
1012 bits of RAM
cycle time: 10-9 sec
The Future
•In near future computers will have
• As many processing elements as our
brain,
• far fewer interconnections, and
• and much faster updates
•Fundamentally different hardware
• Likely requires fundamentally different
algorithms!
• Very much an open question.
What is Artificial
Intelligence?
History of AI
• Prehistory: Greek mythology: Talos, man of bronze, created by Hephaestus; Pygmalion brings Galatea (aka Liza Doolittle) to life.
History of AI
• Pre-20th Century: Talking heads were described, and attempts to build them were made between the 13th and 18th century, most conspicuously by Albertus Magnus and Sir Francis Bacon, alchemists. (Alchemists were also interested in creating a “homonculus,” or perfect miniature of a human.
History of AI
• Term robot was first introduced by
Czech playwright Karel Capek in the
1921 play Rossum’s Universal
Robots.
• The play featured machines created
to simulate human beings.
• Robota is the Czech word for “work”
or “forced workers”/“slaves”.
AI History Summary
• 50’s: Early AI programs, including Samuel's checkers
program, Newell & Simon's Logic Theorist, Gelernter's
Geometry Engine
• 60’s: Search and games, formal logic and theorem proving
• 70’s: Robotics, perception, knowledge representation,
expert systems
• 80’s: More expert systems, AI becomes an industry
• 90’s: Rational agents, probabilistic reasoning, machine
learning
• 00’s: Systems integrating many AI methods, machine
learning, reasoning under uncertainty
• 10’s: reasoning (machine learning) using massive amounts
of data
Definitions of AI vary
The exciting new effort to make computers think … machines with minds, in the full and literal sense” (Haugeland 1985)
“The art of creating machines that perform functions that require intelligence when performed by people” (Kurzweil, 1990)
“The study of mental faculties through the use of computational models” (Charniak et al. 1985)
A field of study that seeks to explain and emulate intelligent behavior in terms of computational processes” (Schalkol, 1990)
What is AI?
Systems that think like humans
Systems that think rationally
Systems that act like humans
Systems that act rationally
Artificial Intelligence is
the study of:
1) Systems Acting like Humans
• Turing test: test for intelligent behaviour• Interrogator writes questions and receives
answers
• System providing the answers passes the test if interrogator cannot tell whether the answers come from a person or not
• Necessary components of such a system form major AI sub-disciplines:• Natural language, knowledge
representation, automated reasoning, machine learning
Acting Humanly: The Turing Test
The Turing Test
Problems:
1) Turing test is not reproducible, or amenable to mathematic analysis.
2) What about physical interaction with interrogator and environment?
What would a computer need to
pass the Turing test?• Natural language processing: to
communicate with examiner.
• Knowledge representation: to store and
retrieve information provided before or during
interrogation.
• Automated reasoning: to use the stored
information to answer questions and to draw
new conclusions.
• Machine learning: to adapt to new
circumstances and to detect and extrapolate
patterns.
2) Systems Thinking like Humans
• Formulate a theory of mind/brain
• Express the theory in a computer
program
• Two Approaches
• Cognitive Science and Psychology
(testing/ predicting responses of
human subjects)
• Cognitive Neuroscience (observing
neurological data)
Thinking Humanly: Cognitive Science
• 1960’s “Cognitive Revolution”: information-processing
psychology replaced behaviorism
• Cognitive science brings together theories and
experimental evidence to model internal activities of
the brain
• What level of abstraction? “Knowledge” or
“Circuits”?
• How to validate models?
• Predicting/testing behavior of human subjects (top-down)
• Direct identification from neurological data (bottom-up)
• Building computer/machine simulated models and
reproduce results (simulation)
3) Systems Thinking Rationally
• “Rational” = ideal intelligence
(contrast with human intelligence)
• Rational thinking governed by precise “laws of thought”
• syllogisms
• notation and logic
• Systems (in theory) can solve problems using such laws
Thinking Rationally: Laws of Thought
• Aristotle (~ 450 B.C.) attempted to
codify “right thinking”
What are correct arguments/thought
processes?
• E.g., “Socrates is a man, all men are
mortal; therefore Socrates is mortal”
• Several Greek schools developed
various forms of logic:notation plus rules of derivation for thoughts.
(led to propositional and predicate logic)
Thinking Rationally: Laws of Thought
Problems:
1)Uncertainty: Not all facts are certain
(e.g., the flight might be delayed).
2)Resource limitations:
• Not enough time to compute/process
• Insufficient memory/disk/etc.
• Etc.
4) Systems Acting Rationally
• Building systems that carry out
actions to achieve the best outcome
• Rational behaviour
• May or may not involve rational
thinking
• i.e., consider reflex action
This is the most commonly adopted
definition
Rational Behavior
Getting computers to do the right
thing based on their circumstances
and what they know.
• No presuppositions about how they should
be designed to do the right thing
• not limited to how people do it
• Evaluation is based on performance, not on
how the task is performed
How to Achieve AI?• How is AI research done?
• AI research has both theoretical and experimentalsides. The experimental side has both basic and applied aspects.
• There are two main lines of research:• One is biological, based on the idea that since humans are
intelligent, AI should study humans and imitate their psychology or physiology.
• The other is phenomenal, based on studying and formalizing common sense facts about the world and the problems that the world presents to the achievement of goals.
• The two approaches interact to some extent, and both should eventually succeed. “It is a race, but both racers seem to be walking.” (John McCarthy)
Applied Areas of AI
• Game play
• Speech and language processing
• Expert reasoning
• Planning and scheduling
• Vision
• Robotics
• Autonomous Vehicles
• Machine Learning (lots of theory here too)
Foundations of AI
• Philosophy: logic, mind, knowledge
• Mathematics: proof, computability,
probability
• Economics: maximizing payoffs
• Neuroscience: brain and neurons
• Psychology: thought, perception, action
• Control Theory: stable feedback systems
• Linguistics: knowledge representation,
syntax
• Data Science: knowledge synthesis
Main Areas of AI Knowledge
representation (including formal logic)
Search, especially heuristic search (puzzles, games)
Planning
Reasoning under uncertainty, including probabilistic reasoning
Learning
Agent architectures
Robotics and perception
Natural language processing
Search
Knowledgerep.
Planning
Reasoning
Learning
Agent
Robotics
Perception
Naturallanguage
... ExpertSystems
Constraintsatisfaction
AI State of the Art
• Which of the following been achieved?• World-class chess playing
• Playing table tennis
• Cross-country driving
• Solving mathematical problems
• Discover and prove mathematical theories
• Engage in a meaningful conversation
• Understand spoken language
• Observe and understand human emotions
• Express emotions
• …
AI State of the Art
• Which of the following been achieved?• World-class chess playing
• Playing table tennis
• Cross-country driving
• Solving mathematical problems
• Discover and prove mathematical theories
• Engage in a meaningful conversation
• Understand spoken language
• Observe and understand human emotions
• Express emotions
• Have emotions
State of the Art
“I could feel – I could smell – a new kind of intelligence across the table”
- Gary Kasparov
State of the Art
“Saying Deep Blue doesn’t really think about chess is like saying an airplane doesn’t really fly because it doesn’t flap its wings.”
– Drew McDermott
Mathematical Calculation
“The Wolfram Language has state-of-the-art capabilities for
the construction, training and deployment of neural network
machine learning systems. Many standard layer types are
available and are assembled symbolically into a network,
which can then immediately be trained and deployed on
available CPUs and GPUs.”
Speech Recognition
Speech Recognition
Repair Scheduling
Autonomous Vehicles
Limits of AI Today
• Today’s successful AI systems
• operate in well-defined domains
• employ narrow, specialize knowledge
• Common-sense Knowledge
• needed in complex, open-ended worlds
• Your kitchen vs. GM factory floor
• understand unconstrained natural language
How do we understand?
• John gave Pete a book.
• John gave Pete a hard time.
• John gave Pete a black eye.
• John gave in.
• John gave up.
• John gave it a go.
• John gave a good accounting of himself.
• John’s legs gave out.
• Give it up for John.
• It is 300 miles, give or take a few.
Can Machines Act/Think Intelligently?
Yes, if intelligence is narrowly defined as
information processing
AI has made impressive achievements
showing that tasks initially assumed to
require intelligence can be automated
Probably not, if intelligence is not
separated from the rest of “being
human”
Major Issues In AI
• How to represent knowledge about the world?
• How to react to new perceived events?
• How to integrate new percepts to past experience?
• How to understand the user?
• How to optimize balance between user goals &
environment constraints?
• How to use reasoning to decide on the best course of
action?
• How to communicate back with the user?
• How to plan ahead?
• How to learn from experience?
Some Big Open Questions
• AI (especially, the “rational agent” approach) assumes that intelligent behaviors are only based on information processing? Is this a valid assumption?
• If yes, can the human brain machinery solve problems that are inherently intractable for computers?
• In a human being, where is the interface between “intelligence” and the rest of “human nature”, e.g.:• How does intelligence relate to emotions felt?
• What does it mean for a human to “feel” that he/she understands something?
• Is this interface critical to intelligence? Can there exist a general theory of intelligence independent of human beings?
Asimov’s I, Robot
In the movie I, Robot, the most impressive feature of the robots is not their ability to solve complex problems, but how they blend human-like reasoning with other key aspects of human beings (especially, self-consciousness, fear of dying, distinction between right and wrong).
AI – A CommentDespite all these achievements, one of the major
philosophers of Cognitive Science wrote:
“… the failure of artificial intelligence to produce
successful simulation of routine commonsense
cognitive competences is notorious, not to say
scandalous. We still don't have the fabled machine
that can make breakfast without burning down the
house; or the one that can translate everyday English
into everyday Italian, or the one that can summarize
texts..”
(Jerry Fodor, The Mind doesn’t Work that Way, 2000)
AI in Business
“Our current AI state is that of a toddler. It can
understand what it sees, what it hears, and then tell you
what it is seeing. This is just the early stage for AI, but
for businesses to really gain value from it, the next step
will be the ability to make more connections and
reasoning about the connections between objects. We
are progressing down a path from going from a toddler
to a slightly more capable learning machine.” Todd Anglin, VP Product Strategy, Progress Software, Nov. 28,
2017
What is a Robot?
Narrow Definition
• “A robot is a reprogrammable
multifunctional manipulator designed to
move material, parts, tools, or
specialized devices through variable
programmed motions for the
performance of a variety of tasks.”
From the Robot Institute of America,1980.
Broader Definition
“A mobile device that
operates with some
degree of autonomy,
usually under computer
control.”
Other Definitions
• An efficient, insensitive person.
• A person who works mechanically
without original thought,
especially one who responds
automatically to the commands of
others.
• A machine that looks like a
human being.
Two Main Design Goals
• Mobility
• Main goal is transport
• Manipulation
• Main goal is to perform an
action on the environment.
Robotics Applications
• Industrial
• Dominant from commercial standpoint
• Manufacturing
• Service
• Everything not industrial
• Telerobots (master/slave systems)
• Virtual Reality Interfaces
Mobile Robot Examples
• Search and Rescue
• Remote-controlled
• Robust, Tethered
Center for Robot-Assisted Search and Rescue
Mobile Robot Examples
• Guides (museum,
tourist attractions,
etc.)Rhino (CMU/Bonn ICS)
Museum Tour-Guide Robots
Rhino, 1997 Minerva, 1998
“How Intelligent Is Minerva?”
• Amoeba?
• Fish?
• Dog?
• Monkey?
• Human?
“How Intelligent Is Minerva?”
fish dog monkey humanamoeba
5.7%
29.5%
25.4%
36.9%
2.5%
“Is Minerva Alive?"
undecided noyes
3.2%
27.0%
69.8%
Mobile Robot Examples
• Walking & running robots
(one or more legs)
• Purpose:
• Handle uneven terrain
• Help us understand
biological locomotionSprawlita (Stanford)
Mobile Robot Examples
• Hopping robots
use accurate
dynamic models
of the system.
• Specialized
controllers
stabilize the
system.
3D Biped (MIT)
Mobile Robot Examples
• Soccer-playing robots
• Cooperating Agents
CMU RoboCup Entry
LegoRobots Game Board
“Bob”
“Jaundice 5”
“Le Saboteur”
“Scooby Doo”
“Psycha...”
RoboCup Challenge
Design a team of robots
that can play soccer!
Manipulator Examples
• Medical robots
(tele-operated)
Manipulator Examples
• Space shuttle arm
(tele-operated)
Manipulator Examples
• Painting robots
• Interesting issues in
coverage, path
planning
Manipulator Examples
• Industrial robots (tele-
operated &
autonomous)
Manipulator Examples
• Baking
Robots!?!
Block Grabber
Sensor Details
Are the following robots?
• Automobile?
• Washing Machine?
• Treadmill?
• Cockroach with implanted control
chip?
• A haptic device?
• A battlebot?
2005 DARPA Challenge
DARPA Grand Challenge
• Race of autonomous vehicles across
California desert
• Vehicles are given a route as series of GPS
waypoints
• But they must intelligently avoid obstacles
and stay on the road
• About 130 miles of dirt roads, off-road,
normal roads, bridges, tunnels, etc.
• Must complete in less than 10 hours
2007 ChallengeThe DARPA Urban Challenge featured autonomous
ground vehicles executing simulated military supply
missions safely and effectively in a mock urban area. Safe
operation in traffic is essential to U.S. military plans to use
autonomous ground vehicles to conduct important
missions.
DARPA awarded prizes for the top three autonomous
ground vehicles that compete in the final event where they
must safely complete a 60-mile urban area course in
fewer than six hours. First prize was $2 million, second
prize was $500,000 and third prize was $250,000. To
succeed, vehicles had to autonomously obey traffic laws
while merging into moving traffic, navigating traffic circles,
negotiating busy intersections and avoiding obstacles.
Honda Asimo Humanoid Robot
Walk
Turn
Stairshttp://world.honda.com/robot/
Sony AIBO
http://www.aibo.com
2009-2011 Urban Challenge• Vehicle must be built upon a full-size stock chassis.
• Vehicle must be capable of loading a mission description
file (MDF) via a standard USB 2.0 flash drive. The vehicle will
have 5 minutes to process a MDF before attempting the
course. The MDF will include information such as accessible
roads, all areas in which the vehicle may travel, stop sign
locations, nominal lane width, maximum and minimum speed
limits, lane markings, and parking spot locations. The MDF
has no implied start or end points. Road blockages will not be
indicated in the MDF.
• To complete the requirements for the Urban Challenge,
each vehicle will complete multiple missions over a defined
route. A mission is a series of checkpoint locations that must
be passed over sequentially by the vehicle. The path between
checkpoints is not specified.
DARPA Urban Challenge
2012 DARPA Grand ChallengeIn the DARPA Robotics Challenge, robots competed with each other
performing disaster response operations in representative scenarios that
could include the following sequence of events:
• Drive a utility vehicle at the site.
• Travel dismounted across rubble.
• Remove debris blocking an entryway.
• Open a door and enter a building.
• Climb an industrial ladder
and traverse an industrial
walkway.
• Use a tool to break through
a concrete panel.
• Locate and close a valve
near a leaking pipe.
• Replace a component
such as a cooling pump.
2015 DARPA Grand Challenge
The DRC Finals competition challenged participating robotics teams and
their robots to complete a difficult course of eight tasks relevant to disaster
response, among them driving alone, walking through rubble, tripping circuit
breakers, turning valves and climbing stairs.
DARPA Subterranean Challenge
(2017-Present)
Current technologies fail to provide persistent situational awareness of the
diverse subterranean operating environments, including tunnels, urban
underground, and cave networks. DARPA encourages participation from
multidisciplinary teams from around the world to address the autonomy,
perception, networking and mobility technologies necessary to map
subsurface networks in unpredictable conditions. To attract a broader range
of participants, the SubT Challenge includes both a physical Systems
competition, as well as a software-only Virtual competition.
Ethics and Societal Impact
• Should we (or can we) make robots
that are autonomously intelligent?
• Should we create autonomously
self-replicating robots?
• How should humans interact with
robots?
• Should humans be replaced by
robots for certain tasks?
Asimov’s 3 (4) Laws of Robotics
• From a 1940’s short story (“I Robot”)
1. Robots must never harm human beings.
2. Robots must follow instructions from humans without violating Rule 1.
3. Robots must protect themselves without violating the other rules.
• Asimov’s robots added Rule 0: Robots must act to prevent harm to humans
(A Few) Issues in Robot Design
• Kinematics
• Dynamics
• Design
• Controls
• Special Areas of Focus:
• Path Planning
• Computer Vision
• Haptics
• Medical Robotics
• Artificial Intelligence