CS325 Artificial Intelligence Robotics I Autonomous Robots (Ch. 25) - - PowerPoint PPT Presentation

▶
cs325 artificial intelligence robotics i autonomous
SMART_READER_LITE
LIVE PREVIEW

CS325 Artificial Intelligence Robotics I Autonomous Robots (Ch. 25) - - PowerPoint PPT Presentation

Apr 13, 2023 222 likes •912 views

CS325 Artificial Intelligence Robotics I Autonomous Robots (Ch. 25) Dr. Cengiz Gnay, Emory Univ. Gnay Robotics I Autonomous Robots (Ch. 25) Spring 2013 1 / 15 Spring 2013 Robots As Killers? The word robot coined by Czech

slide-1
SLIDE 1

CS325 Artificial Intelligence Robotics I – Autonomous Robots (Ch. 25)

  • Dr. Cengiz Günay, Emory Univ.

Spring 2013

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 1 / 15

slide-2
SLIDE 2

Robots As Killers?

The word “robot” coined by Czech writers Capek bros

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 2 / 15

slide-3
SLIDE 3

Robots As Killers?

The word “robot” coined by Czech writers Capek bros Isaac Asimov developed the concept of robotics and three laws:

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 2 / 15

slide-4
SLIDE 4

Robots As Killers?

The word “robot” coined by Czech writers Capek bros Isaac Asimov developed the concept of robotics and three laws:

1

A robot may not injure or cause indirect harm to a human.

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 2 / 15

slide-5
SLIDE 5

Robots As Killers?

The word “robot” coined by Czech writers Capek bros Isaac Asimov developed the concept of robotics and three laws:

1

A robot may not injure or cause indirect harm to a human.

2

It must obey orders except when in conflict with law #1.

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 2 / 15

slide-6
SLIDE 6

Robots As Killers?

The word “robot” coined by Czech writers Capek bros Isaac Asimov developed the concept of robotics and three laws:

1

A robot may not injure or cause indirect harm to a human.

2

It must obey orders except when in conflict with law #1.

3

It must stay alive as long as not in conflict with laws #1 and #2.

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 2 / 15

slide-7
SLIDE 7

Robots As Killers?

The word “robot” coined by Czech writers Capek bros Isaac Asimov developed the concept of robotics and three laws:

1

A robot may not injure or cause indirect harm to a human.

2

It must obey orders except when in conflict with law #1.

3

It must stay alive as long as not in conflict with laws #1 and #2.

Fiction always liked to depict robots taking over

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 2 / 15

slide-8
SLIDE 8

. . . Or As Helpers?

In reality, first we need to make the robots

  • Dr. Thrun says we will soon

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 3 / 15

slide-9
SLIDE 9

. . . Or As Helpers?

In reality, first we need to make the robots

  • Dr. Thrun says we will soon

They can help with?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 3 / 15

slide-10
SLIDE 10

. . . Or As Helpers?

In reality, first we need to make the robots

  • Dr. Thrun says we will soon

They can help with?

Disabled people Children Risky tasks Mundane tasks

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 3 / 15

slide-11
SLIDE 11

. . . Or As Helpers?

In reality, first we need to make the robots

  • Dr. Thrun says we will soon

They can help with?

Disabled people Children Risky tasks Mundane tasks

We’ll focus on the the self-driving car in two lectures

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 3 / 15

slide-12
SLIDE 12

Entry/Exit Surveys

Exit survey: Computer Vision III – Structure from Motion

What additional piece of information an SfM algorithm needs when the objects in the scene also moves? What parameters an SfM algorithm cannot recover?

Entry survey: Robotics I – Autonomous Robots (0.25 pts)

What methods that we have previously seen in this class would be involved in robotics? Name a useful task that you think would be possible to assign to robots.

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 4 / 15

slide-13
SLIDE 13

Self-Driving Cars and DARPA Challenge

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 5 / 15

slide-14
SLIDE 14

Self-Driving Cars and DARPA Challenge

1st DARPA challenge was a failure: cars completed at most 5%. Undergrads like you made Stanley win!

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 5 / 15

slide-15
SLIDE 15

Urban Challenge

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 6 / 15

slide-16
SLIDE 16

Urban Challenge

Google car self-drove 100,000 miles already! We will focus on machine learning, particle filters, and planning.

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 6 / 15

slide-17
SLIDE 17

Robot as an Agent

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 7 / 15

slide-18
SLIDE 18

Robot as an Agent

Is it:

1 Part.-observable? 2 Stochastic? 3 Adversarial? 4 Continuous? 5 Single/Multi? Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 7 / 15

slide-19
SLIDE 19

Robot as an Agent

Is it:

1 Part.-observable 2 Stochastic 3 Adversarial? 4 Continuous 5 Single/Multi? Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 7 / 15

slide-20
SLIDE 20

Perception to Estimate Internal State: Kinematic

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-21
SLIDE 21

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we??

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-22
SLIDE 22

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we?? Roomba is cleaning a room:

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-23
SLIDE 23

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we?? Roomba is cleaning a room: How many dimensions we need for kinematic state?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-24
SLIDE 24

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we?? Roomba is cleaning a room: How many dimensions we need for kinematic state? x, y

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-25
SLIDE 25

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we?? Roomba is cleaning a room: How many dimensions we need for kinematic state? x, y, heading angle Total: 3

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-26
SLIDE 26

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we?? How about for Junior? How many dimensions we need for kinematic state?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-27
SLIDE 27

Perception to Estimate Internal State: Kinematic

Kinematic state: Where in the world are we?? How about for Junior? How many dimensions we need for kinematic state? SAME: x, y, heading angle Total: 3

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 8 / 15

slide-28
SLIDE 28

Including Movement: Dynamic State

Kinematic state: Where in the world are we??

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-29
SLIDE 29

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior:

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-30
SLIDE 30

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior:

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-31
SLIDE 31

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior: Dynamic state: Where are you going??

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-32
SLIDE 32

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior: Dynamic state: Where are you going?? (also includes the kinematic state).

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-33
SLIDE 33

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior: Dynamic state: Where are you going?? (also includes the kinematic state). How many dimensions in dynamic state of Junior?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-34
SLIDE 34

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior: Dynamic state: Where are you going?? (also includes the kinematic state). How many dimensions in dynamic state of Junior? 3 from kinematic forward velocity, v yaw rate: turning angle

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-35
SLIDE 35

Including Movement: Dynamic State

Kinematic state: Where in the world are we?? Junior: Dynamic state: Where are you going?? (also includes the kinematic state). How many dimensions in dynamic state of Junior? 3 from kinematic forward velocity, v yaw rate: turning angle Total: 5

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 9 / 15

slide-36
SLIDE 36

More Dimensions: Flying

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 10 / 15

slide-37
SLIDE 37

More Dimensions: Flying

More quadcopter videos: Aggressive Maneuvers I: State estimation Aggressive Maneuvers II: Hoops! Aggressive Maneuvers III: Trajectory planning Fails!

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 10 / 15

slide-38
SLIDE 38

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-39
SLIDE 39

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-40
SLIDE 40

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-41
SLIDE 41

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-42
SLIDE 42

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6 Dimensions in dynamic state?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-43
SLIDE 43

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6 Dimensions in dynamic state? 6 from kinematic

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-44
SLIDE 44

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6 Dimensions in dynamic state? 6 from kinematic 3 for each dimensional velocity

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-45
SLIDE 45

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6 Dimensions in dynamic state? 6 from kinematic 3 for each dimensional velocity 3 for each angular velocity

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-46
SLIDE 46

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6 Dimensions in dynamic state? 6 from kinematic 3 for each dimensional velocity 3 for each angular velocity Total: 12

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-47
SLIDE 47

Kinematic & Dynamic State of Copters?

Quadcopters: Dimensions in kinematic state? 3D location: x, y, z 3D angles: heading, incline, roll Total: 6 Dimensions in dynamic state? 6 from kinematic 3 for each dimensional velocity 3 for each angular velocity Total: 12 Unlike a car, this can go in all directions!

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 11 / 15

slide-48
SLIDE 48

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-49
SLIDE 49

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot Robotic arm:

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-50
SLIDE 50

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot Robotic arm: Kinematic dimensions:

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-51
SLIDE 51

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot Robotic arm: Kinematic dimensions: 6?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-52
SLIDE 52

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot Robotic arm: Kinematic dimensions: 6? base angles (2) joint angles (2) arm rotation (1), grab (1)

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-53
SLIDE 53

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot Robotic arm: Kinematic dimensions: 6? base angles (2) joint angles (2) arm rotation (1), grab (1) Dynamic dimensions?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-54
SLIDE 54

Kinematic & Dynamic State of Jointed Robots

Honda’s Asimo: a humanoid bipedal robot Robotic arm: Kinematic dimensions: 6? base angles (2) joint angles (2) arm rotation (1), grab (1) Dynamic dimensions? 2 × 6

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 12 / 15

slide-55
SLIDE 55

Localization

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 13 / 15

slide-56
SLIDE 56

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-57
SLIDE 57

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw)

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-58
SLIDE 58

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) Each particle:   x y θ   Remember: estimation and pre- diction?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-59
SLIDE 59

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) Each particle:   x y θ   Remember: estimation and pre- diction? State estimation after ∆t: x′ = x +

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-60
SLIDE 60

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) Each particle:   x y θ   Remember: estimation and pre- diction? State estimation after ∆t: x′ = x + ∆t v cos θ

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-61
SLIDE 61

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) Each particle:   x y θ   Remember: estimation and pre- diction? State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-62
SLIDE 62

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) Each particle:   x y θ   Remember: estimation and pre- diction? State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ θ′ = θ + ∆t w

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-63
SLIDE 63

Monte Carlo Localization: Particle Filter

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) Each particle:   x y θ   Remember: estimation and pre- diction? State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ θ′ = θ + ∆t w 1st approx., but works well.

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 14 / 15

slide-64
SLIDE 64

Localization Question

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ θ′ = θ + ∆t w

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 15 / 15

slide-65
SLIDE 65

Localization Question

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ θ′ = θ + ∆t w Initial state: x = 24, y = 18, θ = 0 v = 5/sec, w =

π 8 sec

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 15 / 15

slide-66
SLIDE 66

Localization Question

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ θ′ = θ + ∆t w Initial state: x = 24, y = 18, θ = 0 v = 5/sec, w =

π 8 sec

Estimate after ∆t = 1 sec?

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 15 / 15

slide-67
SLIDE 67

Localization Question

Roomba: Kinematic state variables: x, y: location θ: heading angle Dynamic state variables: v: forward velocity w: angular velocity (yaw) State estimation after ∆t: x′ = x + ∆t v cos θ y′ = y + ∆t v sin θ θ′ = θ + ∆t w Initial state: x = 24, y = 18, θ = 0 v = 5/sec, w =

π 8 sec

Estimate after ∆t = 1 sec? x′ = 24 + 1 × 5 × 1 = 29 y′ = 18 + 1 × 5 × 0 = 18 θ′ = 0 + 1 × π 8 = π 8

Günay Robotics I – Autonomous Robots (Ch. 25) Spring 2013 15 / 15