Introduction to Robotics

Texts

Primary:
- M. Spong, S. Hutchinson, and M. Vidyasagar, “Robot
Modeling and Control”, Wiley

Secondary:
- Reza N. Jazar, “Theory of Applied Robotics”, Springer, 2010.
- Peter Corke, “Robotics, Vision and Control”, Springer.
- B. Siciliano, L. Sciavicco, L. Villani, G. Oriolo, “Robotics:
Modelling, Planning and Control”, Springer

Software:
- Matlab / Simulink R2012a.
- Robotics Toolbox (use with “Robotics, Vision and Control”)

2
 Course outline
• Main: traditional analysis of robotic manipulators
– Homogeneous transforms
– Forward/inverse kinematics Homework will
– Velocity kinematics, dynamics follow along with
– Motion planning these topics
– Control

• Extras: introduction to special topics

– Sensors and actuators
– Motion planning and control Active research
– Trajectory control for mobile robot areas, potential
– Path tracking subjects for final
– Computer vision projects

Robots in everyday use and popular culture

http://www.robotuprising.com/

• 100s of movies
• Chances are, something you eat, wear, or was made by a robot
4
 Common applications

• Industrial
– Robotic assembly
• Commercial
– Household chores
• Military
• Medical
– Robot-assisted surgery

Common applications

• Planetary Exploration
– Fast, Cheap, and Out of Control
– Mars rover
• Undersea exploration

JHUROV; Johns Hopkins

6
 Industrial robots

• High precision and repetitive tasks

– Pick and place, painting, etc
• Hazardous environments

Representations

• For the majority of this class, we will consider robotic manipulators

as open or closed chains of links and joints
– Two types of joints: revolute (θ) and prismatic (d)

8
 Mathematical modeling of robots
• End-effector/Tool
– Device that is in direct contact with the environment. Usually
very task-specific
• Configuration
– Complete specification of every point on a manipulator
– Set of all possible configurations is the configuration space
– For rigid links, it is sufficient to specify the configuration space
by the joint angles q = [ q1 q2 ... qn ]T
• State space
– Current configuration (joint positions q) and velocities q
• Work space
– The reachable space the tool can achieve
• Reachable workspace
• Dextrous workspace

Example end-effector: Grippers

• Anthropomorphic or task-specific
– Force control v. position control

Utah MIT hand

10
 Common configurations: wrists

• Many manipulators will be a sequential chain of links and joints

forming the ‘arm’ with multiple DOFs concentrated at the ‘wrist’

11

Common configurations: elbow manipulator

• Anthropomorphic arm: ABB IRB1400
• Very similar to the lab arm (RRR)

12
 Workspace: elbow manipulator

13

Common configurations: Stanford arm (RRP)

• Spherical manipulator (workspace forms a set of concentric
spheres)

14
 Common configurations: SCARA (RRP)

Adept Cobra Smart600

SCARA robot

15

Common configurations: cylindrical robot (RPP)

Seiko RT3300 Robot

16
Common configurations: Cartesian robot (PPP)

• Increased structural rigidity, higher precision

– Pick and place operations

17

Workspace comparison

(a) spherical
(b) SCARA
(c) cylindrical
(d) Cartesian

18
 Parallel manipulators
• Some of the links will form a closed chain with ground
• Advantages: Motors can be proximal: less powerful, higher
bandwidth, easier to control
• Disadvantages: Generally less motion, kinematics can be
challenging

6DOF Stewart platform ABB IRB6400 ABB IRB940 Tricept

19

Simple example:
control of a 2DOF planar manipulator
• Move from ‘home’ position and follow the path AB with a constant
contact force F all using visual feedback

20
 Coordinate frames & forward kinematics

• Three coordinate frames: 0 1 2

• Positions:
⎡ x1 ⎤ ⎡ a1 cos (θ1 ) ⎤
⎢ y ⎥ = ⎢ a sin θ ⎥
⎣ 1⎦ ⎣ 1 ( 1)⎦ 2

⎡ x2 ⎤ ⎡ a1 cos (θ1 ) + a2 cos (θ1 + θ 2 ) ⎤ ⎡ x ⎤

⎢ y ⎥ = ⎢ a sin θ + a sin θ + θ ⎥ ≡ ⎢ y ⎥
⎣ 2 ⎦ ⎣ 1 ( 1 ) 2 ( 1 2 ) ⎦ ⎣ ⎦t

• Orientation of the tool frame: 0 1

⎡1 ⎤ ⎡0 ⎤
xˆ0 = ⎢ ⎥ , yˆ 0 = ⎢ ⎥
⎣0 ⎦ ⎣1 ⎦
⎡cos(θ1 + θ 2 ) ⎤ ⎡ − sin(θ1 + θ 2 ) ⎤
xˆ2 = ⎢ ⎥ , yˆ = ⎢ cos(θ + θ ) ⎥
⎣ sin(θ1 + θ 2 ) ⎦
2
⎣ 1 2 ⎦

⎡ xˆ ⋅ xˆ yˆ 2 ⋅ xˆ0 ⎤ ⎡cos(θ1 + θ 2 ) − sin(θ1 + θ 2 ) ⎤

R20 = ⎢ 2 0 ⎥ = ⎢ sin(θ + θ ) cos(θ + θ ) ⎥
⎣ 2 0
xˆ ⋅ yˆ yˆ 2 ⋅ yˆ 0⎦ ⎣ 1 2 1 2 ⎦
21

Inverse kinematics

• Find the joint angles for a desired tool position

xt2 + yt2 − a12 − a22
cos(θ 2 ) = ≡ D ⇒ sin(θ 2 ) = ± 1 − D 2
2a1a2
⎛ 1 − D2 ⎞ ⎛ y⎞ ⎛ a2 sin(θ 2 ) ⎞
−1
θ 2 = tan ⎜ ± ⎟ θ1 = tan −1 ⎜ ⎟ − tan −1 ⎜ ⎟
⎜ D ⎟ ⎝x⎠ ⎝ a1 + a2 cos(θ 2 ) ⎠
⎝ ⎠

• Two solutions!: elbow up and elbow down 22

 Velocity kinematics: the Jacobian
• State space includes velocity

⎡ x2 ⎤ ⎡ −a1 sin(θ1 )θ1 − a2 sin(θ1 + θ 2 )(θ1 + θ2 ) ⎤

⎢ y ⎥ = ⎢ ⎥
⎣ 2 ⎦ ⎣ a1 cos(θ1 )θ1 + a2 cos(θ1 + θ 2 )(θ1 + θ2 ) ⎦
⎡ −a1 sin(θ1 ) − a2 sin(θ1 + θ 2 ) −a2 sin(θ1 + θ 2 ) ⎤ ⎡θ1 ⎤
=⎢ ⎥⎢ ⎥
⎣ a1 cos(θ1 ) + a2 cos(θ1 + θ 2 ) a2 cos(θ1 + θ 2 ) ⎦ ⎣θ2 ⎦
G
= Jq

• Inverse of Jacobian gives the joint velocities:

G G
q = J −1 x
1 ⎡ a2 cos(θ1 + θ 2 ) a2 sin(θ1 + θ 2 ) ⎤ ⎡ x ⎤
=
a1a2 sin(θ 2 ) ⎢⎣ − a1 cos(θ1 ) − a2 cos(θ1 + θ 2 ) −a1 sin(θ1 ) − a1 sin(θ1 + θ 2 ) ⎥⎦ ⎢⎣ y ⎥⎦

• This inverse does not exist when θ2 = 0 or π, called singular

configuration or singularity. 23

Path planning

• In general, move tool from position A to position B while avoiding

singularities and collisions
– This generates a path in the work space which can be used to
solve for joint angles as a function of time (usually polynomials)
– Many methods: e.g. potential fields

• Can apply to mobile agents or a manipulator configuration

24
 Joint control

• Once a path is generated, we can create a desired tool path/velocity

– Use inverse kinematics and Jacobian to create desired joint
trajectories

desired trajectory error controller system dynamics

measured trajectory (w/ sensor noise) actual trajectory

25

Other control methods

• Force control or impedance control (or a hybrid of both)

– Requires force/torque sensor on the end-effector
• Visual servoing
– Using visual cues to attain local or global pose information
• Common controller architectures:
– PID
– Adaptive
– Repetitive
• Challenges:
– Underactuation
– Nonholonomy (mobile agents)
– nonlinearity

26
 General multivariable control overview

joint
manipulator
controllers motor
dynamics dynamics

desired state
joint estimation
torques sensors

inverse estimated
kinematics, configuration
Jacobian
desired
trajectory

27

Sensors and actuators

• Sensors
– Motor encoders (internal)
– Inertial Measurement Units
– Vision (external) Basic quantities for both:
– Contact and force sensors
• Bandwidth
• Sotors/actuators
• Dynamic range
– Electromagnetic
– Pneumatic/hydraulic • Sensitivity
– Electroactive
• Electrostatic
• Piezoelectric

28
 Mobile Agents and SLAM
• For untethered and autonomous applications, the performance of
a robotic system (P), whether defined by capability or robustness,
can be related to an agent’s intelligence (I), mobility (M), and
multiplicity (N):
P ∝ I ⋅ M ⋅ f (N ),
∀I ∈ (0,1), M ∈ (0,1), N ∈ {0,1,2,...}, f : N → n ∈ R +

MicroBat; UCLA
Whegs; CWRU RHex; Michigan

29

Mobile Agents and SLAM

• Monte Carlo localization

• SLAM: ‘Simultaneous locomotion and Map-building’

30
 Computer Vision
• Simplest form: estimating the position and orientation of
yourself or object in your environment using visual cues
– Usually a statistical process
– Ex: finding lines using the Hough space

• More complex: guess the objects in environment

• Biomimetic computer vision: how do animals
accomplish these tasks:
– Obstacle avoidance
• Optical flow?
– Object recognition
• Template matching?
31

MEMS and Microrobotics

• Difficult definition(s):
– Robotic systems with feature sizes < 1mm
– Robotic systems dominated by micro-scale
physics
• MEMS: Micro ElectroMechanical Systems
– Modified IC processes to use ‘silicon as a
mechanical material’

Fearing; Berkeley Donald; Dartmouth Pister; Berkeley

32
 Surgical robotics

• Minimally invasive surgery

– Minimize trauma to the patient
– Potentially increase surgeon’s
capabilities
– Force feedback necessary,
tactile feedback desirable

33

Biomimetic Robots
• Using biological principles to reduce design space

BigDog; Boston Dynamics MFI; Harvard & Berkeley Ayers; Northeastern

34