The Hierarchical Paradigm - City University of New York

Download Report

Transcript The Hierarchical Paradigm - City University of New York

1
From Teleoperation to Autonomy
• Define Intelligent Robot
History
-AI
-Engineering
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
• Be able to describe at least two differences between AI
and engineering approaches to robotics
• Be able to describe the difference between telepresence
and semi-autonomous control
• Have some feel for the history and societal impact of
robotics
Introduction to AI Robotics (MIT Press)
Chapter 1
1
1
History
-AI
-Engineering
Teleop
Case Studies
Programming
Summary
Review
Intelligent Robot
• Mechanical creature which can function
autonomously
– Mechanical= built, constructed
– Creature= think of it as an entity with its own
motivation, decision making processes
– Function autonomously= can sense, act, maybe even
reason; doesn’t just do the same thing over and over
like automation
Introduction to AI Robotics (MIT Press)
Chapter 1
2
1
What are Robots?
• Autonomous mechanical creatures
– Capek 1921: R.U.R.
History
-AI
•
-Engineering
Teleop
Case Studies
Programming •
Summary
Review
Intelligent because teleoperation doesn’t
work, doesn’t scale
Physically situated, but now software
agents or softbots
– Principles from robotics influenced AI
community, esp. planning
– Combines programming, networks,
operating systems, algorithms, …
everything about CS into a system (the
ultimate software engineering project)
www.fradulent.org/rur.htm
Introduction to AI Robotics (MIT Press)
Chapter 1
3
1
Robots Constantly in the Press
History
-AI
-Engineering
Teleop
Case Studies
Programming
Summary
Review
www.sony.com
courtesy of MIT AI Lab
www.irobot.com
courtesy of Honda
Introduction to AI Robotics (MIT Press)
Chapter 1
4
1
Less Famous Cousins at
WTC
Inuktun microTracks
½ iRobot PackBot
Introduction to AI Robotics (MIT Press)
Chapter 1
5
1
Why Robots?
Dirty, Dangerous, Dull Tasks
• JV2010, TRADOC, JFCOM, all branches even down to the
organic level
History
-AI
-Engineering
Teleop
Case Studies
Programming
Summary
Review
– Reconnaissance, MOUT, denial of area, consequence management,
logistics, demining
www.friendlymachines.com
Replace Humans
with Robots
Chapter 1
Introduction to AI Robotics (MIT Press)
6
Why Robots?
Better Than Bio
1
• Robots at WTC…
History
-AI
-Engineering
Teleop
Case Studies
Programming
•
Summary
Review
– voids smaller than person
could enter
– voids on fire or oxygen
depleted
Void:1’x2.5’x60’
NBC Response
– Lose ½ cognitive attention
with each level of protection
• Level A=12.5% of normal
ability
Do Things that Living
Things Can’t
Chapter 1
Introduction to AI Robotics (MIT Press)
Void on fire
7
Major Robot Modalities: UAV,
UGV, UUV
1
•
History
-AI
-Engineering •
Teleop
Case Studies
Programming
Summary
•
Review
Unmanned Aerial Vehicles
– drones since Vietnam: Global Hawk, UCAV
– easy: nothing to hit
– hard: mission sensing, human-in-the-loop control
Unmanned Ground Vehicles
– since 1967
– easy: can always stop and think, a priori maps
– hard: perceiving, e.g., light vegetation vs. wall
Unmanned Underwater Vehicles
– ROVs since 1960s
– easy: run tethers
– hard: platform operation in unfriendly environment
Mobility (platform), Perception,
Communications +HRI, Control (Intelligence), Power
Introduction to AI Robotics (MIT Press)
Chapter 1
8
1
Introduction to AI Robotics (MIT Press)
A Brief History…
Chapter 1
9
1
History
-AI
-Engineering
Teleop
Case Studies
Programming
Summary
Review
Industrial Manipulators
• “Tommy” type of robots: deaf, dumb, and blind
• High precision, fast repetition
• Usually no sensing of the environment
– Welding can be off by an inch…
Introduction to AI Robotics (MIT Press)
Chapter 1
10
1
3 Ways of Controlling a Robot
• “RC-ing”
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
– you control the robot
– you can view the robot and it’s relationship to the
environment
– ex. radio controlled cars, bomb robots
– operator isn’t removed from scene, not very safe
Introduction to AI Robotics (MIT Press)
Chapter 1
11
1
3 Ways of Controlling a Robot
• teleoperation
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
– you control the robot
– you can only view the environment through the robot’s eyes
– don’t have to figure out AI
Introduction to AI Robotics (MIT Press)
Chapter 1
12
1
3 Ways of Controlling a Robot
• semi- or full autonomy
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
–
–
–
–
you might control the robot sometimes
you can only view the environment through the robot’s eyes
ex. Sojouner with different modes
human doesn’t have to do everything
Introduction to AI Robotics (MIT Press)
Chapter 1
13
Components of a Telesystem
(after Uttal 89)
1
• Local
– display
– Local control device
• Communication
• Remote
–
–
–
–
sensor
mobility
effector
power
Remote
Local
Display
Communication
Sensor
Mobility
Control
Effector
Power
Introduction to AI Robotics (MIT Press)
Chapter 1
14
1
Example
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
Remote
Local
Introduction to AI Robotics (MIT Press)
Chapter 1
15
1
Typical Run
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
Introduction to AI Robotics (MIT Press)
Chapter 1
16
1
Problems That You Saw
• no feedback, couldn’t really tell that the robot was stuck
but finally got free
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
– robot doesn’t have “proprioception” or internal sensing to tell
you what the flippers were doing. No crunching noises, no
pose widget to show the flippers
• no localization, mapping-> no idea how far traveled
• partial solution: better instrumentation (but can’t do dead
reckoning well)
– operator doesn’t have an external viewpoint to show itself
relative to the environment
• solution: two robots, one to spot the other
• communications dropout, even though ~3 meters away
• lighting conditions went from dark to very bright
– hard for computer vision or human to adjust
Introduction to AI Robotics (MIT Press)
Chapter 1
17
1
But good for unmodeled events
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
Introduction to AI Robotics (MIT Press)
Chapter 1
18
1
Communications is Important:
DarkStar+7 seconds=DarkSpot
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
• 7 second communications lag (satellite relay)
• “interruption” lag on part of operator
Introduction to AI Robotics (MIT Press)
Chapter 1
19
Predator:
~7:1 human to robot ratio
1
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming •
Summary
Review
Leo’s unofficial
Predator page
4 people to control it (52-56 weeks of training)
– one for flying
– two for instruments
– one for landing/takeoff
• plus maintenance, sensor processing and routing
• lack of self-awareness– in Kosovo, come along side in helicopter
and shoot down
Introduction to AI Robotics (MIT Press)
Chapter 1
20
1
Summary of Teleop Problems
• cognitive fatigue
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
• communications dropout
• communications bandwidth
• communications lag
• too many people to run one robot (hidden cost)
Introduction to AI Robotics (MIT Press)
Chapter 1
21
1
Telesystems Best Suited For:
• the tasks are unstructured and not repetitive
• the task workspace cannot be engineered to permit the
History
use of industrial manipulators
Teleop
-Motivation • key portions of the task require dexterous manipulation,
-Components
especially hand-eye coordination, but not continuously
-Problems
-Alternatives • key portions of the task require object recognition or
Case Studies
situational awareness
Programming
• the needs of the display technology do not exceed the
Summary
Review
limitations of the communication link (bandwidth, time
delays)
• the availability of trained personnel is not an issue
Introduction to AI Robotics (MIT Press)
Chapter 1
22
1
Teleop Improvements:
Telepresence
• Telepresence
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
– improves human control, reduces simulator sickness
and cognitive fatigue by providing sensory feedback
to the point that teleoperator feels they are “present”
in robot’s environment
– increases demands on bandwidth
Introduction to AI Robotics (MIT Press)
Chapter 1
23
1
Teleop Improvements:
Supervisory Control
• Semi-autonomous
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
– Supervisory Control
• human is involved, but routine or “safe” portions of the
task are handled autonomously by the robot
• is really a type of mixed-initiative
• Shared Control/ Guarded Control
– human initiates action, interacts with remote by adding
perceptual inputs or feedback, and interrupts execution as
needed
– robot may “protect” itself by not bumping into things
• Traded Control
– human initiates action, does not interact
Introduction to AI Robotics (MIT Press)
Chapter 1
24
1
Teleop Improvements:
Mixed-Initiative
• Levels of Initiative
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
– do only what told to do (teleoperation)
– recommend or augment (cognitive
augmentation)
– act and report
– act on own and supervise itself (autonomy)
Introduction to AI Robotics (MIT Press)
Chapter 1
25
1
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
“No Hands Across America”
• 1994
• CMU NavLab
• Pittsburgh to San Diego
– 2897 miles total
– 2849 autonomously
• Autonomous or MixedInitiative?
Introduction to AI Robotics (MIT Press)
Chapter 1
26
1
Mixed-Initiative Matching Game
Application
Level of Initiative
History
Teleop
-Motivation
-Components
-Problems
-Alternatives
Case Studies
Programming
Summary
Review
• teleoperation
• cognitive augmentation
• act and report
• act on own
Introduction to AI Robotics (MIT Press)
• janitorial robot
• medical robot
(telemedicine)
• high flying surveillance
drone
• combat aerial vehicle
• resupply system for
bringing water to fire
fighters
• guard dog robot
• “nurse” robot
Chapter 1
27
1
Collaborative Teleoperation
1
mpg: June 2, 2000 SRDR Miami Beach: view from Inuktun as it falls
Urban is stuck, Inuktun can’t help from
current perspective
1. Driven off 3rd floor
2. Hoisted to 2nd floor by tether
3. Has better view, changing
configuration & rocking extend view
Introduction to AI Robotics (MIT Press)
3
mpg: June 2, 2000 SRDR Miami Beach: view from Inuktun from hoisted position
2
still: June 2, 2000 SRDR Miami Beach
Chapter 1
28
1
2000 AAAI Mobile Robot
History
Teleop
Case Studies
Programming
Summary
Review
• 2 robots helping each other reduced collision errors,
sped up time navigating confined space, righting
Introduction to AI Robotics (MIT Press)
Chapter 1
29
Example:
Mixed-Initiative & Collab. Teleop
1
•
•
History
Teleop
Case Studies
Programming
Summary
Review
9/2000 DARPA Tactical
Mobile Robots demonstration
Robot used an intelligent
assistant agent to look for signs
of snipers hiding in urban
rubble
–
–
–
–
•
•
motion
skin color
difference in color
thermal (IR camera)
Human navigated mother robot
using viewpoint of 2nd robot
(not in picture)
Once deposited the human
moved the daughter robot, and
either saw a sniper or was
alerted by the agent
Introduction to AI Robotics (MIT Press)
Chapter 1
30
1
•
•
History
Teleop
•
Case Studies
Programming •
Summary
Review
•
•
•
AI provides the “other stuff”
knowledge representation
understanding natural langugage
learning
planning and problem solving
inference
search
vision
Introduction to AI Robotics (MIT Press)
Chapter 1
31
1
Example User Expectation of AI
• Proposed Goal: 1:1 soldier:any robot, where 1 soldier is
responsible for 1 or more active robots but does not have to pay
continuous attention to them.
Young Frankenstein
4 specialists: 1 specialist: 1 specialist: 1 specialist:
1 specialist 1 soldier
1 vehicle
1 vehicle
n vehicles 1 modality
UAVs as
theater
assets
Introduction to AI Robotics (MIT Press)
MAVs as
organic
assets
Flocks
of MAVs
Chapter 1
Field reconfigurable
UUVs
MAV-UGV
cooperative
monitoring
32
1
More Reasonable Expections
Consolidation
History
Teleop
Case Studies
Programming
Summary
Review
agents with
“tactical”
autonomy,
toolkits
Vehicle success is
still based on human,
but robot is “in front”
Dedicated Autonomy Systems
Mass-produced
dedicated
agents
Cooperating
“pack” or “herd”
agents
Human intermittent
attention as team
coordinator, not
with individuals
Reconfigurable AutonomySystems
Fieldreconfigurable
agents
Introduction to AI Robotics (MIT Press)
Chapter 1
Human primary
responsibility as a
tool builder, expert
advisor. Peer-level
communication
33
1
Programming Notes
• You always need telesystem or human intervention as a
backup
History
Teleop
Case Studies
Programming
Summary
Review
– at some point a human will need to take control
– embed in your design
• “Roboticists automate what is easy and leave the rest to
the human”- Don Norman
• The user interface is absolutely critical
– User interface make up 60% of commercial code
– Useful= is the program purpose useful?
• usually given to designer via specifications and requirements
– Usable= can a human use it efficiently?
• designer must conduct usability studies
• avoid “if I can use it, some one else will”
Introduction to AI Robotics (MIT Press)
Chapter 1
34
Example of How an “Internal”
Display Can Hurt
1
History
Teleop
Case Studies
Programming
Summary
Review
iRobot PackBot
video, FLIR, 2 way audio
•
•
•
gamer joystick plus laptop with video & audio
robot state: battery, comms, orientation, camera, encoders
was not used on rubble pile at WTC because it scared off rescuers: too
complicated, too long to boot, too toy
– now integrated with Land Warrior– used in Afghanistan
Introduction to AI Robotics (MIT Press)
Chapter 1
35
1
Summary
• Teleoperation arose a partial solution to
autonomy
History
Teleop
Case Studies
Programming
Summary
Review
– cognitive fatigue, high comms bandwidth, long
delays, and many:one human to robot ratios
– Telepresence tries to reduce cognitive fatigue
through enhanced immersive environments
– Semi-autonomy tries to reduce fatigue, bandwidth by
delegating portions of the task to robot
• mixed-initiative
• Teleop isn’t simple and improvements aren’t
just “better user interfaces”
Introduction to AI Robotics (MIT Press)
Chapter 1
36
1
Review Questions
• What is an intelligent robot?
• What is the difference between engineering and AI
History
robotics?
Teleop
Case Studies • What are 3 types of control?
Programming
• What are the parts of a telesystem?
Summary
Review
• What are problems with teleoperation?
• What’s the difference between telepresence and semiautonomous control?
• What are the levels of initiative (mixed-initiative)?
• What are alternatives to traditional teleoperation?
Introduction to AI Robotics (MIT Press)
Chapter 1
37