0% found this document useful (0 votes)
242 views71 pages

Lecture Notes For: by Z.X. Li and Y.Q. Wu

an Open-chain [ ml θ̈ ] + [ −ml sθ cθ φ̇ ] + [ mgl sθ ] = [ 0 ] Manipulator Newton-Euler 2 2 Equations Coordinate- This document summarizes robot dynamics and control topics including Lagrangian equations, inertial properties of rigid bodies, dynamics of open-chain manipulators, Newton-Euler equations, and Lagrange's equations with constraints. It provides mathematical formulations and examples for dynamics of simple and spherical pendulums using Lagrangian mechanics. invariant algorithms for robot dynamics Lagrange’s Equations with Constraints

Uploaded by

Abdullah Demirel
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
242 views71 pages

Lecture Notes For: by Z.X. Li and Y.Q. Wu

an Open-chain [ ml θ̈ ] + [ −ml sθ cθ φ̇ ] + [ mgl sθ ] = [ 0 ] Manipulator Newton-Euler 2 2 Equations Coordinate- This document summarizes robot dynamics and control topics including Lagrangian equations, inertial properties of rigid bodies, dynamics of open-chain manipulators, Newton-Euler equations, and Lagrange's equations with constraints. It provides mathematical formulations and examples for dynamics of simple and spherical pendulums using Lagrangian mechanics. invariant algorithms for robot dynamics Lagrange’s Equations with Constraints

Uploaded by

Abdullah Demirel
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 71

Chapter 4 Robot Dynamics and Control

Summer School-Math. Methods in [email protected] 13-31 July 2009


1

Chapter 4
Lecture Notes
Robot
Dynamics and
Control
for
Lagrangian A Mathematical Introduction to
Equations

Inertial Robotic Manipulation


Properties of
Rigid Body

Dynamics of
an Open-chain By
Z.X. Li∗ and Y.Q. Wu#
Manipulator

Newton-Euler
Equations

Coordinate- Dept. of ECE, Hong Kong University of Science & Technology
invariant #
algorithms for
School of ME, Shanghai Jiaotong University
robot
dynamics

Lagrange’s
24 July 2009
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

Summer School-Math. Methods in [email protected] 13-31 July 2009


2

Chapter 4
Robot
Dynamics and
Chapter 4 Robot Dynamics and Control
Control
1 Lagrangian Equations
Lagrangian
Equations

Inertial 2 Inertial Properties of Rigid Body


Properties of
Rigid Body

Dynamics of
3 Dynamics of an Open-chain Manipulator
an Open-chain
Manipulator

Newton-Euler
4 Newton-Euler Equations
Equations

Coordinate- 5 Coordinate-invariant algorithms for robot dynamics


invariant
algorithms for
robot
dynamics 6 Lagrange’s Equations with Constraints
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.1 Lagrangian Equations


y
3
◻ A Simple Example:
Fy F
Chapter 4
m Fx
Robot
Dynamics and
Control
mg
Lagrangian ♢ Review: x
Equations Newton’s Equation: Lagrangian Equation:
Inertial
d ∂L ∂L
Properties of
− = Fx
Rigid Body mẍ = Fx dt ∂ẋ ∂x
Dynamics of
mÿ = Fy − mg
⇔ d ∂L ∂L
an Open-chain
Manipulator − = Fy
dt ∂ẏ ∂y
Newton-Euler Momentum: Px = mẋ
Equations Lagrangian function:
Py = mẏ
Coordinate-
d d
L = T − V, Px = ∂L
∂ẋ , Py =
∂L
∂ẏ
dt Px = Fx , dt Py = Fy − mg
invariant
algorithms for Kinetic energy:
robot
dynamics T = 21 m(ẋ2 + ẏ2 )2
Lagrange’s
Equations with
Potential energy:
Constraints V = mgy
Chapter 4 Robot Dynamics and Control

4.1 Lagrangian Equations


4
◻ Generalization to multibody systems:
Chapter 4 qi , i = 1, . . . , n: generalized coordinates
Robot
Dynamics and
y Kinetic energy:
Control
m3
q3 T = T(q, q̇)
Lagrangian
Equations Potential energy:
Inertial V = V(q)
Properties of m2
Rigid Body Lagrangian:
Dynamics of q2 L(q, q̇) = T(q, q̇) − V(q)
an Open-chain
Manipulator
m1
τi , i = 1, . . . , n: external force on qi
Newton-Euler Lagrangian Equation:
Equations
q1
Coordinate- x
invariant d ∂L ∂L
algorithms for − = τi , i = 1, . . . , n
robot dt ∂q̇i ∂qi
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.1 Lagrangian Equations


5
◇ Example: Pendulum equation
Generalized coordinate: y
Chapter 4 1
Robot θ∈S
Dynamics and
Control Kinematics: x
Lagrangian x = l sin θ, y = −l cos θ
⇒ ẋ = l cos θ ⋅ θ̇, ẏ = l sin θ ⋅ θ̇
Equations

Inertial
Properties of Kinetic energy: θ
T(θ, θ̇) = m(ẋ + ẏ ) = ml θ̇
Rigid Body
1 2 2 1 2 2
Dynamics of
an Open-chain 2 2
Manipulator Potential energy:
Newton-Euler
Equations
V = mgl(1 − cos θ) mg
Coordinate-
Lagrangian function:
L = T − V = ml2 θ̇ − mgl(1 − cos θ), ⇒ = −mgl sin θ
invariant 1 ∂L ∂L
algorithms for = ml2 θ̇,
robot
dynamics
2 ∂ θ̇ ∂θ
Equation of motion:
− = τ ⇒ ml2 θ̈ + mgl sin θ = τ
Lagrange’s
Equations with d ∂L ∂L
Constraints
dt ˙ ∂θ
Chapter 4 Robot Dynamics and Control

4.1 Lagrangian Equations


6
◇ Example: Dynamics of a Spherical Pendulum
Chapter 4

⎡ l sin θ cos ϕ ⎤
Robot


Dynamics and

r(θ, ϕ) = ⎢ l sin θ sin ϕ ⎥


Control

⎢ −l cos θ ⎥
⎣ ⎦
Lagrangian
Equations

T = m∥ṙ∥2 = ml2 (θ̇ 2 + (1 − cos2 θ)ϕ̇2 )


Inertial 1 1
Properties of
Rigid Body 2 2 θ
Dynamics of
an Open-chain
V = −mgl cos θ

L(q, q̇) = ml2 (θ̇ + (1 − cosθ )ϕ̇2 ) + mgl cos θ


Manipulator
1
Newton-Euler
Equations 2
Coordinate-
invariant ϕ mg
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.1 Lagrangian Equations


7




Chapter 4

⎪ = (ml2 θ̇) = ml2 θ̈, = ml2 sin θ cos θ ϕ̇2 − mgl sin θ
d ∂L d ∂L

Robot


⎪ dt ∂ θ̇ dt
Dynamics and


Control ∂θ


⎪ = (ml2 sin2 θ ϕ̇) = ml2 sin2 θ ϕ̈ + 2ml2 sin θ cos θ θ̇ ϕ̇,
d ∂L d ∂L

Lagrangian



=0
Equations

Inertial ⎩ dt ∂ ϕ̇ dt ∂ϕ
Properties of

θ̈ ] + [ −ml2 sθ cθ ϕ̇2 ] + [ mglsθ ] = [ 0 ]


Rigid Body

[ 2 2 ][
Dynamics of
ml2 0
an Open-chain 0 ml sθ ϕ̈ 2ml2 sθ cθ θ̇ ϕ̇ 0 0
Manipulator

Newton-Euler
Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


8
◻ Kinetic energy of a rigid body:
Chapter 4
Robot
Dynamics and r
Control B
Lagrangian
Equations
ra
Inertial
Properties of
Rigid Body A gab
Dynamics of
an Open-chain
Manipulator

Newton-Euler
Equations
Volume occupied by the body: V
Coordinate-
Mass density: ρ(r)
invariant
algorithms for Mass: m= ∫
V
ρ(r)dV

r≜
robot
1
dynamics

Lagrange’s
Mass center:
m V ∫
ρ(r)rdV
Equations with Relative to frame at the mass center: r=0
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


9
In A-frame
Chapter 4

ρ(r)∥ṗ + Ṙr∥2 dV = ρ(r)(∥ṗ∥2 + 2ṗT Ṙr + ∥Ṙr∥2 )dV


Robot
1 1
Dynamics and
Control T=
2 V ∫ ∫ 2 V
= m∥ṗ∥2 + ṗT Ṙ ρ(r)rdV +
Lagrangian
1 1
Equations
∫ ∫ ρ(r)∥Ṙr∥2 dV
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Inertial 2 V 2
=0
Properties of
Rigid Body

Dynamics of
an Open-chain
Manipulator
1
Newton-Euler
Equations 2 ∫ V
ρ(r)∥Ṙr∥2 dV
1 1 1
∫ ρ(r)∥R Ṙr∥ dV = 2 ∫ ρ(r)∥ω̂r∥ dV = 2 ∫ ρ(r)∥r̂ω∥ dV
Coordinate- T 2 2 2
invariant =
algorithms for 2

∫ ρ(r)(−ω r̂ ω)dV = 2 ω ( ∫ (−ρ(r)r̂) dV) ω ≜ 2 ω I ω


robot
dynamics 1 1 T 2 1 T 2 T
=
Lagrange’s 2
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


10


where

⎡ Ixx Ixy Ixz


Chapter 4
I =− ∫ ρ(r)r̂2 dV ≜ ⎢
⎢ Ixy Iyy Iyz ⎥

Robot

⎣ ⎦
Dynamics and ⎢ Ixz Iyz Izz
Control
with
∫ ρ(r)(y + z2 )dxdydz, Ixy = − ∫ ρ(r)xydxdydz
Lagrangian
2
Equations Ixx =
Inertial
Properties of
Rigid Body

T = m∥ṗ∥2 + (ωb )T I b ωb = m∥RT ṗ∥2 + (ωb )T I b ωb


1 1 1 1
Dynamics of
an Open-chain
Manipulator
2 2 2 2
= (V ) [ ] Vb
1 b T mI 0
Newton-Euler
2 0 Ib
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶
Equations

Coordinate-
invariant Mb
algorithms for

For V̂ b = g −1 ⋅ ġ, with ω̂b = RT ⋅ Ṙ and vb = RT ṗ, M b is the


robot
dynamics

Lagrange’s
Equations with
Generalized inertia matrix in B frame.
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


11
◇ Example: M b for a rectangular object
m z
Chapter 4
Robot ρ=
lωh
ρ(y2 + z2 )dxdydz
Dynamics and
Control

Lagrangian
Ixx = ∫ V y
Equations h ω l
x
∫ ∫ ∫ (y2 + z2 )dxdydz
2 2 2
Inertial =ρ h l
Properties of
Rigid Body
− h2 − ω2 − 2l

= ρ( (lω3 h + lωh3 )) = (ω2 + h2 )


Dynamics of 1 m w
an Open-chain
Manipulator 12 12
h ω l

Ixy = − ∫ ρxydv = −ρ ∫ ∫ ∫
Newton-Euler 2 2 2
xydxdydz
− h2 − ω2 − 2l
Equations
V
Coordinate-
invariant l
h ω

= −ρ x ∣ dydz = 0
algorithms for y 2 2
robot
dynamics ∫ ∫ 2

− h2 − ω2
2

2 −l
Lagrange’s 2
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


12

12 (w +h ) ⎥
⎡ m 2 2

⎢ 0 0
12 (l + h2 ) ⎥ , M = [ mI03×3 I0 ]

2 ⎥
m 2
I =⎢ 0 0

12 (w + l ) ⎦


⎢ m 2
Chapter 4 0 0
Robot
Dynamics and
g2 (t)
Control
g0
Lagrangian
Equations
V̂1 = g1−1 ⋅ ġ1 , M1 A B
1
Inertial
Properties of T = V1T M1 V1
Rigid Body 2 g1 (t) g0
Dynamics of V1 = Adg0 V2
an Open-chain
Manipulator

T = (Adg0 V2 )T M1 (Adg0 V2 ) = V2T AdTg0 M1 Adg0 V2 ≜ V2T M2 V2


Newton-Euler
Equations
1 1 1
Coordinate- 2 2 2
invariant
algorithms for ◻ M under change of frames:
robot
dynamics

Lagrange’s M2 = AdTg0 M1 Adg0


Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


13
◇ Example: Dynamics of a 2-dof planar robot

Chapter 4


Robot ⎡ Ixxi 0 0
⎥ , i = 1, 2
Dynamics and ⎢

Ii = ⎢ 0 Iyyi 0
⎣ ⎦
Control
⎢ 0 0 Izzi
Lagrangian
Equations

Inertial

T(θ, θ̇) = m1 ∥v1 ∥2 + ωT1 I1 ω1


Properties of l2
Rigid Body 1 1
Dynamics of 2 2 r 2 θ2

+ m2 ∥v2 ∥ + ωT2 I2 ω2
an Open-chain
Manipulator
1 2 1
l1
Newton-Euler 2 2 y
Equations r1
θ1
Coordinate- x
invariant
0 0
ω1 = [ 0 ] ω2 = [ ]
algorithms for
0
θ̇ 1 + θ̇ 2
robot
dynamics
θ̇ 1
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.2 Intertial Properties of Rigid Body


14
xi
Pi = [ yi ]: Mass center γi : Distance from joint i to mass center
0
ẋ1 = −r1 s1 θ̇ 1
Chapter 4
Robot x1 = r1 c1
Dynamics and ⇒
Control
y1 = r1 s1 ẏ1 = r1 c1 θ̇ 1
x2 = l1 c1 + r2 c12 ẋ2 = −(l1 s1 + r2 s12 )θ̇ 1 − r2 s12 θ̇ 2
Lagrangian
Equations

y2 = l1 s1 + r2 s12

ẏ1 = (l1 c1 + r2 c12 )θ̇ 1 + r2 c12 θ̇ 2
Inertial
Properties of

T(θ, θ̇) = m1 (ẋ1 + ẏ1 ) + Iz1 θ̇ 12 + m2 (ẋ2 + ẏ2 ) + Iz2 (θ̇ 12 + θ̇ 22 )


Rigid Body
1 2 2 1 1 2 2 1
Dynamics of
an Open-chain 2 2 2 2
α + 2βc δ + βc2
= [θ̇ 1 θ̇ 2 ] [ δ + βc 2 ] [ θ̇ 1 ]
Manipulator
1
Newton-Euler
2 δ θ̇ 2
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Equations 2
Coordinate-
M(θ)
Iz1 + Iz2 + m1 r12 + m2 (l12 + r22 ), β
= m2 l1 r2 , δ = Iz2 + m2 r22 , L = T
invariant
algorithms for
α=
−βs2 θ̇ 2 −βx
robot

⇒ M(θ) [ θ̈ 1 ] + [ ] [ θ̇ 1 ] = [ ττ21 ]
dynamics

Lagrange’s
Equations with θ̈ 2 βs2 θ̇ 1 0 θ̇ 2
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


15
◻ Dynamics of open-chain manipulator:
Definition:
Li : frame at mass center of link i, gsli (θ) = expξ̂1 θ 1 ⋯ expξ̂i θ i gsli (o)
Chapter 4
Robot
Dynamics and
Control θ1
Lagrangian θ2 θ3
Equations l1 l2
L2 L3
Inertial
Properties of
Rigid Body r1 r2
l0
Dynamics of L1
an Open-chain
Manipulator r0
S
Newton-Euler

⎡ ⎤
Equations

⎢ θ̇ 1 ⎥
⎢ ⋮ ⎥
Coordinate-

⎢ ⎥
invariant

⎢ ⎥
algorithms for

Vsli = Jsli (θ)θ̇ = [ξ1 ξ2 ⋯ ξi 0 ⋯ 0] ⎢ θ̇ i ⎥ = Ji (θ)θ̇


robot

⎢ ⎥
b b † † †

⎢ ⎥
dynamics
θ̇ i+1
⎢ ⋮ ⎥
⎢ ⎥
Lagrange’s

⎣ ⎦
Equations with
Constraints θ̇ n
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


16
ξj† = Ad−1 (eξ̂j+1 θ j+1 ⋯eξ̂i θ i gsli (0))ξj , j ≤ i

Ti (θ, θ̇) = (Vslbi )T Mib Vslbi = θ̇ T JiT (θ)Mib Ji (θ)θ̇


Chapter 4 1 1
Robot
Dynamics and 2 2
n
T(θ) = ∑ Ti (θ, θ̇) = θ̇ T M(θ)θ̇,
Control
1
Lagrangian
Equations i=1 2
1 n
M(θ) = ∑ JiT (θ)Mib Ji (θ) = ∑ Mij (θ)θ̇ i θ̇ j
Inertial
Properties of
Rigid Body
i 2 i,j=1
hi (θ): Height of Li , Vi (θ) = mi ghi (θ), V(θ) = ∑ mi ghi (θ)
Dynamics of
an Open-chain
Manipulator
i=1
Newton-Euler
Equations
Lagrange’s Equation:


Coordinate- d ∂L ∂L
invariant = Γi , i = 1, . . . , n,
algorithms for dt ∂ θ̇ i ∂θ i
d ⎛n ⎞ n
robot

∑ Mij θ̇ j = ∑ Mij θ̈ j + Ṁij θ̇ j


dynamics
d ∂L
dt ∂ θ̇ i dt ⎝ j=1 ⎠ j=1
Lagrange’s =
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


17
∂L 1 n ∂Mkj ∂Mij
θ̇ k θ̇ j −
∂V
= ∑ Ṁij = ∑ θ̇ k
∂θ i 2 j,k=1 ∂θ i ∂θ i k ∂θ k
Chapter 4
Robot n n ∂Mij 1 ∂Mkj
⇒ ∑ Mij θ̈ j + ∑ ( θ̇ j θ̇ k − θ̇ k θ̇ j ) +
Dynamics and ∂V
Control = Γi
j=1 j,k=1 ∂θ k 2 ∂θ i ∂θ i
Lagrangian
Equations n n
⇒ ∑ Mij θ̈ j + ∑ Γijk θ̇ k θ̇ j +
∂V
Inertial = Γi
Properties of
Rigid Body j=1 j,k=1 ∂θ i
1 ∂Mij ∂Mik ∂Mkj
( + − )
Dynamics of
an Open-chain Γijk =
Manipulator
2 ∂θ k ∂θ j ∂θ i
Newton-Euler

θ̇ i ⋅ θ̇ j , i ≠ j : Coriolis force
Equations

Coordinate- θ̇ i2 : Centrifugal force


n
1 n ∂Mij ∂Mik ∂Mkj
Define: cij (θ, θ̇) = ∑ Γijk θ̇ k = ∑ ( + − )θ̇ k
invariant
algorithms for
robot
dynamics k=1 2 k=1 ∂θ k ∂θ j ∂θ i
Lagrange’s
Equations with ⇒ M(θ)θ̈ + C(θ, θ̇)θ̇ + N(θ) = τ
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


18

Property 1:
M(θ) = M T (θ), θ̇ T M(θ)θ̇ ≥ 0, θ̇ T M(θ)θ̇ = 0 ⇔ θ = 0
Chapter 4
Robot 1

Ṁ − 2C ∈ Rn×n is skew symmetric


Dynamics and
Control
2
Lagrangian
Equations
Proof :
Inertial

(Ṁ − 2C)ij = Ṁij − 2cij (θ)


Properties of
Rigid Body

Dynamics of n ∂M ∂Mij ∂Mkj


θ̇ k − θ̇ k − θ̇ k +
∂Mik
=∑
an Open-chain ij
Manipulator θ̇ k
Newton-Euler k=1 ∂θ k ∂θ k ∂θ j ∂θ i
n ∂Mkj
θ̇ k −
Equations
∂Mik
Coordinate- =∑ θ̇ k
invariant
k=1 ∂θ i ∂θ j
algorithms for

Switching i and j shows (Ṁ − 2C)T = −(Ṁ − 2C)


robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


19
◇ Example: Planar 2-DoF Robot (continued)
Chapter 4 m11 (θ) = α + 2β cos θ 2 , m22 = δ
m12 (θ) = m21 (θ) = δ + β cos θ 2
Robot
Dynamics and
Control

Lagrangian
Equations
c11 (θ, θ̇) = −β sin θ 2 ⋅ θ̇ 2 , c12 (θ, θ̇) = −β sin θ 2 (θ̇ 1 + θ̇ 2 )
Inertial c21 (θ, θ̇) = β sin θ 2 ⋅ θ̇ 1 , c22 (θ, θ̇) = 0
Properties of


⎪ ( + − )=
1 ∂M11 ∂M11 ∂M11 1 ∂M11


Rigid Body
Γ111 =

=0



2 ∂θ 1 ∂θ 1 ∂θ 1 2 ∂θ 1

Dynamics of



an Open-chain
⎪ ( + − )= = −β sin θ 2
1 ∂M11 ∂M12 ∂M21 1 ∂M11


Manipulator
Γ2 =
⎪ 11

Newton-Euler 2 ∂θ 2 ∂θ 1 ∂θ 1 2 ∂θ 2


⎪ Γ12
Equations

⎪ ( + − )= = −β sin θ 2
1 ∂M12 ∂M11 ∂M12 1 ∂M11


1
=


Coordinate-


2 ∂θ 1 ∂θ 2 ∂θ 1 2 ∂θ 2


invariant




algorithms for

( + − )= − = −β sin θ 2
1 ∂M12 ∂M12 ∂M22 ∂M12 1 ∂M22


robot 2

dynamics Γ12 =
2 ∂θ 2 ∂θ 2 ∂θ 1 ∂θ 2 2 ∂θ 1
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


20


⎪ = ( + − )= −
1 ∂M21 ∂M21 ∂M11 ∂M21 1 ∂M11


1

Γ21 = β sin θ 2



2 ∂θ 1 ∂θ 1 ∂θ 2 ∂θ 1 2 ∂θ 2


⎪ 2

Chapter 4

⎪ = ( + − )=
1 ∂M21 ∂M22 ∂M21 1 ∂M22

Robot
Dynamics and

⎪ 21
Γ =0

Control 2 ∂θ 2 ∂θ 1 ∂θ 2 2 ∂θ 1



⎪ = ( + − )=
1 ∂M22 ∂M21 ∂M12 1 ∂M22

Lagrangian


1
=0

Γ22

Equations


2 ∂θ 1 ∂θ 2 ∂θ 2 2 ∂θ 1




Inertial

⎪ = ( + − )=
1 ∂M22 ∂M22 ∂M22 1 ∂M22


Properties of 2

Rigid Body Γ22 =0
2 ∂θ 2 ∂θ 2 ∂θ 2 2 ∂θ 2
−2β sin θ 2 ⋅ θ̇ 2 −β sin θ 2 ⋅ θ̇ 2
Dynamics of
an Open-chain
Ṁ − 2C = [ ]
Manipulator
−β sin θ 2 ⋅ θ̇ 2 0
−2β sin θ 2 ⋅ θ̇ 2 −2β sin θ 2 (θ̇ 1 + θ̇ 2 )
Newton-Euler

−[ ]
Equations

Coordinate- 2β sin θ 2 ⋅ θ̇ 1 0
invariant

β sin θ 2 (2θ̇ 1 + θ̇ 2 )
=[ ] ⇐ skew-symmetric
algorithms for
0
−β sin θ 2 (2θ̇ 1 + θ̇ 2 )
robot
dynamics
0
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


21
◇ Example: Dynamics of a 3-dof robot
Chapter 4 θ1
Robot
Dynamics and θ2 θ3

⎡ ⎤ ⎡ 0 ⎤ ⎡ 0 ⎤
l1 l2
⎢ ⎥ ⎢ −l0 ⎥ ⎢ −l0 ⎥
Control 0
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
L2 L3

⎢ ⎥ ⎢ 0 ⎥ ⎢ l1 ⎥
0
Equations ξ 1 = ⎢ ⎥ , ξ 2 = ⎢ −1 ⎥ , ξ 3 = ⎢ −1 ⎥
Lagrangian 0
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
r1 r2
⎢ ⎥ ⎢ 0 ⎥ ⎢ 0 ⎥
0
⎢ ⎥ ⎢ 1 ⎥ ⎢ ⎥
0 l0
⎣ ⎦ ⎣ ⎦ ⎣ 1 ⎦
Inertial L1
1
Properties of r0
Rigid Body S

Dynamics of
an Open-chain
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
⎢ I ( 0 ) ⎥ ⎢ ( r1 ) ⎥ ⎢ ( l 1 + r2 ) ⎥
0 0 0
gsl1 (0) = ⎢ ⎥, gsl (0) = ⎢ I ⎥, gsl (0) = ⎢ I ⎥
Manipulator
⎢ ⎥ 2 ⎢ ⎥ 3 ⎢ ⎥
⎢ 0 r0 ⎥ ⎢ 0 l0 ⎥ ⎢ 0 l0 ⎥
⎣ ⎦ ⎣ ⎦ ⎣ ⎦
Newton-Euler
Equations
1 1 1
Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


22
⎡ mi 0 0 ⎤
⎢ 0 mi 0 ⎥
⎢ ⎥
⎢ 0 0 mi ⎥
0
Mi = ⎢ ⎢ 0 ⎥ ⎥
Chapter 4

⎢ 0 Iy i 0 ⎥
Robot Ixi 0
⎢ ⎥
⎢ 0 Iz i ⎥
Dynamics and
0
⎣ ⎦
Control

Lagrangian
0
Equations mi : The mass of the object
Inertial Ixi : The moment of inertia about the x axis
Properties of

Γ112 = (Iy2 − Iz2 − m2 r12 )c2 s2 + (Iy2 − Iz3 )c23 s23 − m3 (l1 c2 + r2 c23 )(l1 s2 + r2 s23 )
Rigid Body

Γ113 = (Iy3 − Iz3 )c23 s23 − m3 r2 s23 (l1 c2 + r2 c23 )


Dynamics of
an Open-chain

= (Iy2 − Iz2 − m2 r12 )c2 s2 + (Iy3 − Iz3 )c23 s23 − m3 (l1 c2 + r2 c23 )(l1 s2 + r2 s23 )
Manipulator
1
Γ12
Γ13 = (Iy3 − Iz3 )c23 s23 − m3 r2 s23 (l1 c2 + r2 c23 )
Newton-Euler 1

= (Iz2 − Iy2 + m2 r12 )c2 s2 + (Iz3 − Iy3 )c23 s23 + m3 (l1 c2 + r2 c23 )(l1 s2 + r2 s23 )
Equations
1
Γ21
Γ22 = −l1 m3 r2 s3 , Γ23 = −l1 m3 r2 s3 , Γ23 = −l1 m3 r2 s3
Coordinate-
invariant 3 2 3

Γ311 = (Iz3 − Iy3 )c23 s23 + m3 r2 s23 (l1 c2 + r2 c23 ) , Γ32


algorithms for 2
robot = l 1 m3 r2 s 3
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


23
, V(θ) = m1 gh1 (θ) + m2 gh2 (θ) + m2 gh3 (θ)
∂V
N(θ, θ̇) =
∂θ
Chapter 4

gsli (θ) = eξ̂1 θ 1 ⋯eξ̂i θ i gsli (0) ⇒


Robot
Dynamics and
Control

Lagrangian h1 (θ) = r0 , h2 (θ) = l0 − r1 sin θ, h3 (θ)


= l0 − l1 sin θ 2 − r2 sin(θ 2 + θ 3 )gsli (θ)
Equations

Inertial

= eξ̂1 θ 1 ⋯eξ̂i θ i gsli (0) ⇒


Properties of
Rigid Body

h1 (θ) = r0 , h2 (θ) = l0 − r1 sin θ, h3 (θ)


Dynamics of
an Open-chain

= l0 − l1 sin θ 2 − r2 sin(θ 2 + θ 3 )
Manipulator

Newton-Euler
Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


24
⎡ 0 0 0 ⎤
⎢ 0 0 0 ⎥
⎢ ⎥
⎢ 0 0 0 ⎥
J1 = Jsl1 (θ) = ⎢ 0 0 0 ⎥
⎢ ⎥
Chapter 4 b

⎢ 0 0 0 ⎥
Robot

⎢ 1 0 0 ⎥
Dynamics and

⎣ ⎦
Control

⎡ −r1 c2 0 0 ⎤
Lagrangian

⎢ 0 0 ⎥
⎢ ⎥
Equations

⎢ −r1 0 ⎥
0
J2 = Jslb2 (θ) = ⎢ −1 0 ⎥
Inertial
0
⎢ ⎥
Properties of

⎢ −s2 0 0 ⎥
Rigid Body 0
⎢ 0 0 ⎥
⎣ ⎦
Dynamics of
an Open-chain c2
⎡ −l2 c2 − r2 c23 0 ⎤
⎢ 0 ⎥
Manipulator
0
⎢ ⎥
⎢ −r2 − l1 c3 −r2 ⎥
Newton-Euler
0 l1 s3
J3 = Jsl3 (θ) = ⎢ −1 ⎥
Equations

−1
b 0
⎢ ⎥
⎢ 0 ⎥
0
−s23
Coordinate-

⎢ ⎥
invariant

⎣ 0 ⎦
algorithms for
0
robot c23 0
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


25

M11 M12 M13


Chapter 4 M(θ) = [ M21 M22 M23 ] = J1T M1 J1 + J2T M2 J2 + J3T M3 J3
Robot M31 M32 M33
M11 = Iy2 s2 + Iy3 s223 + Iz1 + Iz2 c22 + Iz3 c223 + m2 r12 c22 + m3 (l1 c2 + r2 c23 )2
Dynamics and
Control 2
Lagrangian
Equations M12 = M13 = M21 = M31 = 0
M22 = Ix2 + Ix3 + m2 l12 + M2 r12 + m3 r22 + 2m3 l1 r2 c3
Inertial
Properties of

M23 = Ix3 + m3 r22 + m3 l1 r2 c3


Rigid Body

Dynamics of

M32 = Ix3 + m3 r22 + m3 l1 r2 c3


an Open-chain
Manipulator

M33 = Ix3 + m3 r22


Newton-Euler
Equations

Coordinate-
n
invariant
1 n ∂Mij ∂Mik ∂Mkj
Cij (θ, θ̇) = ∑ Γijk θ̇ k = ∑( + − ) θ̇ k
algorithms for
robot
dynamics
k=1 2 k=1 ∂θ k ∂θ j ∂θ i
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


26

◻ Additional Properties of the dynamics in


Chapter 4
Robot terms of POE:
Dynamics and
Control Define:

⎪Ad−1ξ̂ j+1 θ j+1 ξ̂ θ i > j
Lagrangian




Equations

⎪ e ⋯e i i
Aij = ⎨I
Inertial



i=j
Properties of



Rigid Body

Dynamics of ⎪
⎩0 i<j
Ji (θ) = Adg −1 (0) [Ai1 ξ1 ⋯Aii ξi 0⋯0]
an Open-chain
Manipulator

Newton-Euler sl i

Mi′ ( intertia of ith link in S)


Equations

Coordinate-
= AdTg−1 Mi Adg −1
sl (0)
i sl (0)
i
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.3 Dynamics of Open-chain Manipulator


27

Property 2:
n
Mij (θ) = ∑ ξiT ATli Ml′ Alj ξj
Chapter 4
Robot
Dynamics and
Control l=max(i,j)

1 n ∂Mij ∂Mik ∂Mkj


Cij (θ, θ̇) = ∑( + − ) θ̇ k
Lagrangian
Equations

Inertial 2 k=1 ∂θ k ∂θ j ∂θ i
Properties of
Rigid Body where
n
= ∑ ([Ak−1,i ξi , ξk ]T ATlk Ml′ Alj ξj
∂Mij
Dynamics of
an Open-chain
Manipulator
∂θ k l=max(i,j)
Newton-Euler
Equations
+ξiT ATli Ml′ Alk [Ak−1,j ξj , ξk ])
Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.4 Newton-Euler Equations


28
◻ Newton-Euler equations in body frame:
Newtons Equation:
Chapter 4
Robot

(mṗ) = mp̈
Dynamics and d
Control f =
Lagrangian
dt
Equations
Spatial angular momentum:
Inertial

I′ ⋅ ωs = R(I ⋅ ωb ) = R ⋅ I ⋅ RT ⋅ωs
Properties of

´¹¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹¶
Rigid Body

Dynamics of
an Open-chain I′
Manipulator

Newton-Euler
Equations r
Coordinate-
T
invariant
algorithms for
robot
dynamics
f
Lagrange’s S
g ∶ (R, p)
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.4 Newton-Euler Equations


29

(I ω ) = (RI b RT ωs ) = I s ω̇s + ṘI b RT ωs + RI b ṘT ωs


d s s d
Chapter 4
τ=
dt dt
= I ω̇ + ṘR I ω − RI b RT ω̂s ωs = I s ω̇s + ωs × (I s ωs )
Robot
Dynamics and s s T s s
±̂
Control

Lagrangian s ω
Equations

Inertial ◻ Transformation of all equations to


Properties of
Rigid Body twist/wrench in body frame:
(mṗ) = (mRvb ) = mṘvb + mRv̇b , RT f s = mRT Ṙvb + mv̇b
Dynamics of
d d
an Open-chain
dt dt
⇒ f = mωb × v̇b + mv̇b ,
Manipulator
b
Newton-Euler
Equations

τb = RT τs = RT (RI b ωb ) = I b ω̇b + ωb × I b ωb
d
Coordinate-
invariant dt
] + [ ωb × mv
algorithms for

⇒[ b ][ ] = [ f b ] = F b (∗)
mI 0 b b b b
robot v̇
ω ×I ω
dynamics b b b
0 I ω̇ τ
Lagrange’s
Equations with
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶ ´¹¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹¶
Constraints
M Vb
Chapter 4 Robot Dynamics and Control

4.4 Newton-Euler Equations


30

Chapter 4 Define: se(3) × se(3) ↦ se(3), ( ξ̂1 , ξ̂2 ) ↦ ξ̂1 ξ̂2 − ξ̂2 ξ̂1 ≜ ξ̂, if
ξ̂i = [ ω̂0i v0i ] , i = 1, 2
Robot
Dynamics and
Control
then
ξ̂ = [ (ω1 ×0ω2 ) ω̂1 v2 −

Lagrangian
Equations ω̂2 v1 ] = ad ⋅ ξ
0 ξi 2
Inertial
where
ad ξ1 = [ ω̂01 ω̂ v̂1 ]
Properties of
Rigid Body

Dynamics of 1
an Open-chain Thus
(∗) ⇔ Mb V̇b − adV b Mb Vb = F b
Manipulator

Newton-Euler
Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.4 Newton-Euler Equations


31
◻ Coordinate invariance of Newton-Euler
equations:
Chapter 4
Robot
Dynamics and r g0
Control
B
M1 V̇1 − adTV1 M1 V1 = F1
g2
Lagrangian
Equations
V1 = Adg0 V2
AdTg0 F1 = F2 ⇒
Inertial
Properties of

F1 = (AdTg0 )−1 F2 = (Ad−1g 0 ) F2


Rigid Body T A g1
Dynamics of

M1 = Ad−T −1
an Open-chain
g0 M2 Adg0
Manipulator

−T −1 −T −1
Newton-Euler
Equations ⇒ Adg0 M2 Adg0 Adg0 V̇2 − adT(Adg V̇2 ) Adg0 M2 Adg0 Adg0 V1 = Ad−T
g 0 F2
0
Coordinate-

Since adAdg0 V = Adg0 adV Ad−1


invariant
algorithms for
robot
g0 , we have
dynamics

Lagrange’s
Equations with
M2 V̇2 − adV2 M2 V2 = F2
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


32
Cn Cn+1
C2
C1

Chapter 4
Robot
Dynamics and
Control

Lagrangian C0
Equations
Ci : Frame fixed to link i, located along the ith axis
Inertial
Properties of Fi : Generalized force link i − 1 exerting on link i, expressed in Ci
Rigid Body τi : Joint torque of link i
Dynamics of gi−1,i : Transformation of Ci relative to Ci−1
gi−1,i (θ i ) = eξ̂i θ i ⋅ gi−1,i (0) = gi−1,i (0)eξ̂i θ i
an Open-chain ′
Manipulator

Newton-Euler ξi = Adg −1 (0) ⋅ ξi′ : ith axis in Ci frame.


⎧ ⎡ ⎤
i−1,i

⎪ ⎢ 0 ⎥
Equations


⎪ ⎢ ⎥∶

⎪ ⎢ ⎥
Coordinate-



⎪⎢ zi ⎥
invariant Revolute joint.
ξ = ⎨⎣ ⎦
algorithms for

⎪ ⎡ ⎤
robot


⎪ ⎢ ⎥
⎪ ⎢ ⎥∶
dynamics
z

⎪ ⎢ ⎥
i

⎪ ⎢ ⎥
Lagrange’s Prismatic joint.
⎩⎣ ⎦
Equations with
Constraints
0
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


33

−1
Chapter 4
Robot
⇒ gi−1,i ⋅ ġi−1,i = ξ̂i ⋅ θ̇ i
Dynamics and
Control
Mi : Moment of inertia in Ci
−mi r̂i
Mi = [ m ir̂ I − m 2 ] I:
Lagrangian mI mi : Mass of link i
Equations
i i i i r̂i i Angular moment of inertia
Inertial
Properties of gi = gi−1 gi−1,i
V̂i = gi−1 ⋅ ġi = gi−1,i
−1
V̂i−1 gi−1,i + ξ̂i θ̇ i
Rigid Body

Dynamics of

Vi = Adg −1 Vi−1 + ξi θ̇ i
an Open-chain
Manipulator
i−1,i
Newton-Euler
−1 −1 −1 ˙
V̂˙ i = ġi−1,i V̂i−1 gi−1,i + gi−1,i V̂i−1 ġi−1,i + gi−1,i V̂i−1 gi−1,i + ξ̂i θ̈ i
Equations

−1 −1 −1 −1
= −gi−1,i V̂i−1 gi−1,i + gi−1,i
Coordinate-
invariant
algorithms for
ġi−1,i gi−1,i V̂i−1 gi−1,i gi−1,i ġi−1,i
+ g −1 V̂˙ g + ξ̂ θ̈
robot
dynamics
i−1,i i−1 i−1,i i i
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


34
= − ξ̂i θ̇ i (Adg −1 Vi−1 )∧ + (Adg −1 Vi−1 )∧ ξ̂i θ̇ i + (Adg −1 V̇i−1 )∧ + ξ̂i θ̈ i
⇒ V̇i = ξi θ̈ i + Adg −1 V̇i−1 − ad ξi θ̇ i (Adg −1 Vi−1 )
i−1,i i−1,i i−1,i

Chapter 4 i−1,i i−1,i


Robot
Dynamics and ◻ Forward Recursion:
i = 0 ∶ V0 = 0, V̇0 = [ 0 ]
Control g
Lagrangian


Equations



⎪ gi−1,i = gi−1,i (0)eξ̂i θ i


Inertial


Properties of

⎨ Vi = Adgi−1,i −1 Vi−1 + ξ i θ̇ i
Rigid Body





Dynamics of


⎩ V̇i = ξi θ̈ i + Adgi−1,i
⎪ −1 V̇i−1 − ad
ξ i θ̇ i (Adgi−1,i
−1 Vi−1 )
an Open-chain
Manipulator

Newton-Euler
Equations

Coordinate- ◻ Backward Recursion:


invariant
Fn+1 : End-effector wrench, gn,n+1 : transform from tool frame to Cn
Fi = AdTg−1 ⋅ Fi+1 + Mi V̇i − adTVi ⋅ Mi Vi
algorithms for
robot
dynamics i,i+1
Lagrange’s
Equations with τi = ξiT ⋅ Fi
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


35

−1 ⋅ V0 + ξ 1 θ̇ 1
Chapter 4
Robot V1 = Adg0,1
Fn = AdTg−1 ⋅ Fn+1 + Mn V̇n − adTVn ⋅ (Mn Vn )
Dynamics and
Control

Lagrangian n,n+1

⎡ θ̇ ⎤
Define:
⎢ 1 ⎥
Equations

V = [ ⋮ ] ∈ R6n×1 , θ̇ = ⎢ ⎥
V1 ξ1 0 0
⎢ ⋮ ⎥∈R ,ξ=[ 0 ⋱ 0 ] ∈ R6n×n
Inertial
n
⎢ θ̇ n ⎥
Properties of
Vn ⎣ ⎦
Rigid Body
0 0 ξn
⎡ Adg −1 ⎤
⎢ ⎥
Dynamics of
F1 τ1 ⎢ ⎥
an Open-chain

F = [ ⋮ ] ∈ R6n×1 , τ = [ ⋮ ] ∈ Rn , P0 = ⎢ 0 ⎥ ∈ R6n×6
0,1

⎢ ⋮ ⎥
Manipulator

Fn τn ⎢ 0 ⎥
⎣ ⎦
Newton-Euler
Equations

Coordinate-
invariant
Pt = [0⋯0 Adg −1 ] ∈ R
T 6×6n
n,n+1
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


36
V1 = Adg0,1
−1 ⋅ V0 + ξ 1 θ̇ 1

V2 − Adg −1 V1 = ξ2 θ̇ 2
Chapter 4 1,2
Robot
Dynamics and ⋮
Control
Vn − Adg −1 Vn−1 = ξn θ̇ n
⎡ 0 ⎤
Lagrangian
n−1,n

⎢ ⎥ ⎡ V1 ⎤
Equations
I 0 ⋯
⎢ −Adg1,2 ⋮ ⎥⎢ V ⎥
⎢ ⎥⎢ 2 ⎥
Inertial I ⋱
⇒⎢ ⎥⎢ ⋮ ⎥
−1
Properties of

⎢ 0 ⋱ ⋱ 0
I ⎥ ⎢ ⎥
⎢ ⎥ ⎣ Vn ⎦
Rigid Body

⎣ ⎦ ´¹¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¶
Dynamics of 0 0 −Adg −1
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ V
n−1,n
an Open-chain
Manipulator
G−1
⎡ Adg −1 ⎤ ⎡ ξ1 ⎤⎡ θ̇ 1 ⎤
⎥⎢ ⎥
Newton-Euler

⎢ ⎥ ⎢ ⎢ ⎥
Equations

⎢ 0 0,1 ⎥ ⎢ ⎥ ⎢ ⎥
=⎢ ⎥V + ⎢ ⎥⎢
⎥⎢ ⋮ ⎥
ξ θ̇
⎢ ⋮ ⎥ 0 ⎢
Coordinate-


2 2
⎢ 0 ⎥ ⎢ ξn ⎥
invariant

⎣ ⎦ ⎣ ⎦⎢ θ̇ n ⎥
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ´¹¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹⎦

algorithms for

´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
robot


dynamics

Lagrange’s P0 ξ
Equations with
θ̇
Constraints Thus V = GP0 V0 + Gξ θ̇
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


37

where
⎡ I ⎤
⎢ Ad −1 ⎥
Chapter 4
0 0 ⋯ 0
⎢ ⎥
Robot

⎢ ⎥
Dynamics and I 0 ⋯ 0
G=⎢ ⎥ ∈ R6n×6n
Control g1,2
⎢ Ad Adg −1 I ⋱ ⋮ ⎥
⎢ ⋮ 1,3 ⎥
Lagrangian g −1
⎢ ⎥
2,3
I
⎢ Adg −1 ⎥
Equations
⋮ ⋱ 0
Inertial
Properties of ⎣ 1,n
Adg −1
2,n
⋯ Adg −1
n−1,n
I ⎦
Rigid Body

Dynamics of
an Open-chain V̇1 = ξ1 θ̈ 1 + Adg −1 V̇0 − ad ξ1 θ̇ 1 (Adg −1 V0 )
0,1 0,1
Manipulator

Newton-Euler

ξ 2 θ̇ 2 (Adg1,2
−1 V1 )
Equations
−1 V̇1 = ξ 2 θ̈ 2 − ad
V̇2 − Adg1,2
Coordinate-
invariant

V̇n − Adg −1 V̇n−1 = ξn θ̈ n − ad ξn θ̇ n (Adg −1 Vn−1 )


algorithms for
robot
dynamics
n−1,n n−1,n
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


38

⎡ ⋯ 0 ⎤ ⎡ V̇ ⎤ ⎡ Adg −1 ⎤ ⎡ ξ1 ⎤⎡ ⎤
⎢ −Ad −1 ⋮ ⎥ ⎥⎢ ⎥
I
⎥⎢ ⎥ ⎢ 0,1 ⎥ ⎢
0
⎢ ⋱ ⎥⎢
θ̈ 1

⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎢ θ̈ 2 ⎥
I 1
⎢ ⎥⎢ ⎥=⎢ ⎥ V̇0 +⎢ ⎥
⎥⎢ ⎥
Chapter 4 g 1,2
⋱ ⋱ ⋱
V̇2 ξ2
⎢ ⋮ ⎥ ⎢ ⋮ ⎥ ⎢ ⎢ ⋮
0
⎢ ⎥ ⎥
I ⎥⎢ ⎥ ⎢ ⎥ ⎢ ξn ⎥
Robot 0 0
⎢ 0 −Adg −1 ⎦⎢ ⎥
⎣ ⎦ ⎣ V̇n ⎦ ⎣ ⎦ ⎣ ⎣ θ̈ n ⎦
Dynamics and 0 0
Control
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
n−1,n
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Lagrangian P0 ξ
G−1
Equations

⎡ −ad ξ θ̇ ⋯ ⎤ ⎡ Ad −1 ⎤
⎢ 0 0 ⎥⎢ ⎥
Inertial
⎢ −ad ξ 2 θ̇ 2 ⋱ ⋮ ⎥⎢ ⎥
⎢ ⎥⎢
Properties of 1 1 g 0,1
+⎢ ⎥⎢ ⎥ V0
0
⋮ ⋱ ⋱ ⋮ ⎥
0
⎢ ⎥⎢
Rigid Body
⎢ ⋯ 0 −ad ξ n θ̇ n ⎥
0 ⎥
Dynamics of ⎣ 0 ⎦⎣ 0 ⎦
an Open-chain
Manipulator
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
ad ξ θ̇
Newton-Euler
Equations

⎡ −ad ξ θ̇ ⋯ ⎤⎡ ⋯ 0 ⎤
⎢ ⎥ ⎢ Ad −1 0 ⋮ ⎥ ⎡ V1 ⎤
0 0

0 0
⎢ ⎥⎢ ⎥⎢ V ⎥
Coordinate-
−ad ξ 2 θ̇ 2 ⋱ ⋮
+⎢ ⎥⎢ ⎥⎢ 2 ⎥
1 1
invariant
⎢ ⎥⎢ 0 ⎢ ⎥ ⎥
0 g 1,2
⎢ ⋮ ⋱ ⋱ ⎥⎢ ⋱ ⋱ ⎢ ⋮
0 ⎥ ⎥
algorithms for 0
⎢ ⎥⎢ ⎥ ⎣ Vn ⎦
0
⎣ ⋯ −ad ξ n θ̇ n ⎦⎣ ⎦
robot
dynamics 0 0 0 0 Adg −1
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
n−1,n

Lagrange’s Γ
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


39

Chapter 4 Thus
Robot
Dynamics and
Control V̇ = G ⋅ ξ θ̈ + G ⋅ P0 V̇0 + G ⋅ ad ξ θ̇ P0 V0 + G ⋅ ad ξ θ̇ ΓV
Lagrangian
Equations
Finally the backward recursion:
Inertial

Fn = AdTg−1 Fn+1 + Mn V̇n − adTVn ⋅ (Mn Vn )


Properties of
Rigid Body

Dynamics of n,n+1

AdTg−1 Fn + Mn−1 V̇n−1 − adTVn−1 ⋅ (Mn−1 Vn−1 )


an Open-chain
Manipulator Fn−1 =
n−1,n
Newton-Euler
Equations ⋮
F1 = AdTg−1 F2 + M1 V̇1 − adTV1 (M1 V1 )
Coordinate-
invariant
algorithms for 1,2
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


40
⎡ ⎤
⎢ I −AdTg−1 0 ⎥ ⎡ ⎤
⎢ 0 ⎥ ⎡ F1 ⎤ ⎡ M 1 ⎤⎢ V̇1 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ ⎥ ⎢ F2 ⎥ ⎢ ⎥⎢ ⎥
1,2
Chapter 4
⇒⎢⎢ 0 I ⋱ 0 ⎥⎢
⎥⎢
⎥=⎢
⎥ ⎢
M2 ⎥⎢
⎥⎢
V̇2 ⎥

⎢ ⋱ −AdTg−1 ⎥⎥⎢⎢ ⋮ ⎥⎥ ⎢⎢ ⋱ ⎥⎢ ⎥
Robot
⎢ ⎥⎢ ⎥
⋮ ⋱ ⋮
⎢ n−1,n ⎥⎣ Fn ⎦ ⎣ ⎦⎢ ⎥
Dynamics and
⎢ ⎥
M ⎣ ⎦
´¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Control
⎣ 0 ⋯ 0 I ⎦
n V̇n
⎡ ⎤
Lagrangian M


0
⎥ ⎡ −adT ⎤
⎢ ⎥ M1
Equations

⎢ ⎥
⎥ Fn+1 + ⎢ ⎥[
V1
+⎢ ][ ⋮ ]
⋮ V1
⎢ T ⎥
Inertial

⎢ ⎥ ⎢ ⎥
Properties of 0 ⋱ ⋱
⎢ Adg −1 ⎥ ⎣ Vn ⎦
Mn Vn
⎣ n,n+1 ⎦
Rigid Body T −ad
Dynamics of
an Open-chain ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
T
Manipulator
PtT adV
Newton-Euler
Equations

F = GT M V̇ + GT PtT Fn+1 +GT ⋅ adTV MV


±
Coordinate-
invariant
algorithms for
robot Ft
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


41

τ = ξT ⋅ F
Chapter 4
Robot
τ = ξ T GT M V̇ + ξ T GT PtT Ft + ξ T GT ⋅ adTV MV
Dynamics and
Control = ξ T GT M(Gξ θ̈ + GP0 V̇0 + G ⋅ ad ξ θ̇ P0 V0 + G ⋅ ad ξ θ̇ ΓV)
Lagrangian ⇒
Equations + ξ T GT PtT Ft + ξ T GT ⋅ adTV MV
= ξ T GT MGξ θ̈ + ξ T GT MGP0 V̇0 + ξ T GT MG ⋅ ad ξ θ̇ ΓV
Inertial
Properties of
Rigid Body

Dynamics of + ξ T GT PtT Ft + ξ T GT ⋅ adTV MV


an Open-chain
Manipulator Finally we get:
³¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ · ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹µ ³¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ·¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ µ
Newton-Euler
M(θ) C(θ,θ̇)

ξ T GT MGξ θ̈ + ξ T GT (MG ⋅ ad ξ θ̇ Γ + adTV M)Gξ θ̇


Equations

Coordinate-
invariant
+ ξ T GT MGP0 V̇0 + ξ T GT PtT Ft = τ
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶ ´¹¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¶
algorithms for
robot

JtT (θ)
dynamics
ϕ(θ)
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.5 Coordinate-invariant algorithms for robot dynamics


42

M(θ)θ̈ + C(θ, θ̇) + ϕ(θ) + JtT (θ)Ft = τ


Chapter 4
Robot
Dynamics and
Control

Lagrangian
Equations
M(θ) = ξ T GT MGξ
Inertial C(θ, θ̇) = ξ T GT (MGad ξ θ̇ Γ + adTV M)Gξ θ̇
Properties of

ϕ(θ) = ξ T GT MGP0 V̇0


Rigid Body

Dynamics of
an Open-chain
Manipulator
Jt = Pt Gξ
Newton-Euler Property 3:
Equations
Γn = 0,
G = (I − Γ)−1 = I + Γ + Γ2 + ⋯ + Γn−1
Coordinate-
invariant
algorithms for
robot
dynamics I + ΓG = G
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


43
Definition: Holonomic constraints
Chapter 4
Given generalized coordinates q = (q1 , . . . , qn ) ∈ E, a holonomic
Robot constraint is a set of constraint equations:
hi (q) = 0, i = 1, . . . , k
Dynamics and
Control

Lagrangian
Equations
Q = h (0) is a manifold of dim n − k ≜ m if the constraints are
−1

linearly independent.
Tq Q ∶ {V ∈ Tq E∣dhi ⋅ V = 0, ∀i = 1, . . . , k} ⊂ Tq E
Inertial
Properties of
Rigid Body
subspace of permissible velocities.
Tq∗ Q⊥ ∶ {f ∈ Tq∗ E∣⟨f , v⟩ = 0, ∀V ∈ Tq Q} = span{dh1 , . . . , dhk }
Dynamics of
an Open-chain
Manipulator

Newton-Euler
subspace of constraint forces.
Equations

Coordinate- Definition: Constraint forces


invariant
∂h T
algorithms for Γ= ∂q ⋅λ
robot
k
dynamics
λ ∈ R is the vector of relative magnitudes of constraint forces.
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


44

Definition: Pfaffian Constraints


Chapter 4
Robot A Pfaffian constraint has the form:
Dynamics and
Control
A(q)q̇ = 0, A(q) ∈ Rk×n
Given a Pfaffian constraint,
∆q = {Vq ∈ Tq E∣A(q) ⋅ Vq = 0} ⊂ Tq E
Lagrangian
Equations

Inertial
Properties of Distribution of permissible velocities.
Rigid Body

Dynamics of
an Open-chain
Manipulator Definition:
Newton-Euler A(q)q̇ = 0 is holonomic (or integrable) iff ∆q is an involutive
Equations
distribution, or iff
∃hi ∶ E ↦ R, i = 1, . . . , k s.t. ∆q = Tq Q, Q = h−1 (0)
Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


45

§ Constraint forces:
Γ = AT (q) ⋅ λ, λ ∈ Rk
Chapter 4
Robot
Dynamics and
Control

Lagrangian
Equations
§ Kinetic energy:
1
Inertial
Properties of
T(q, q̇) = q̇T ⋅ M(q) ⋅ q̇
Rigid Body
2
Dynamics of
an Open-chain § Potential energy:
Manipulator

Newton-Euler V(q)
Equations

Coordinate-
invariant § Lagrangian:
algorithms for
robot
dynamics L(q, q̇) = T(q, q̇) − V(q)
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


46

Chapter 4
Robot
◻ Lagrange’s equations with constraints:
M(q)q̈ + C(q, q̇) + N(q) + AT (q)λ = F
Dynamics and
Control

Lagrangian
Equations

Inertial
Properties of
◻ Explicit solution for constraint forces:
Rigid Body
A(q)q̈ + Ȧ(q)q̇ = 0
(AM −1 AT )λ = AM −1 (F − C − N) + Ȧq̇
Dynamics of
an Open-chain
Manipulator

Newton-Euler
Equations
λ = (AM −1 AT )−1 (AM −1 (F − C − N) + Ȧq̇)
Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


47

◇ Example:
x2 + y2 = l 2
Chapter 4
Robot

[x y] [ ẋẏ ] = 0
Dynamics and

²
Control

Lagrangian
y
Equations A(q)

L(q, q̇) = m(ẋ2 + ẏ2 ) − mgy


Inertial
x 1
Properties of
Rigid Body l 2
[ m0 m ] [ ÿ ] + [ mg ] + [ y ] λ = 0
0 ẍ 0 x
(x, y)
Dynamics of
θ
λ = (AM −1 AT )−1 (AM −1 (F − C − N) + Ȧq̇)
an Open-chain
Manipulator

= − 2 (gy + ẋ2 + ẏ2 )


Newton-Euler
Equations m
Coordinate- mg l
= ∥[ y ] λ∥ = y + (ẋ2 + ẏ2 )
invariant x mg m
algorithms for
robot l l
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


48
◻ Lagrange-d’Alembert formulation:
Given the Pfaffian constraint A(q)q̇ = 0 and virtual displacement
δq ∈ Rk , we have:
Chapter 4
Robot
Dynamics and
Control
Theorem 1 (D’alembert Principle):
Lagrangian
Equations Forces of constraints do no virtual work!
Inertial (AT (q)λ) ⋅ δq = 0 for A(q)δq = 0
Properties of
Rigid Body

Dynamics of
an Open-chain
Manipulator
1717–1783
Newton-Euler
Equations
Theorem 2 (Lagrange-d’Alembert Equation):
(
Coordinate-
invariant d ∂L ∂L
algorithms for − − τ) ⋅ δq = 0, A(q)δq = 0
robot dt ∂q̇ ∂q
dynamics

Lagrange’s
Equations with
Constraints
1736-1813
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


49

Chapter 4 Let A(q) = [A1 (q) A2 (q)], and A2 (q) ∈ Rk×k is invertible, then
δq1 ∈ Rn−k are free variables:
Robot
Dynamics and
Control

Lagrangian
Equations
δq2 = −A−1
2 (q)A1 (q)δq1

⇒(
Inertial d ∂L ∂L
Properties of − − τ) ⋅ δq
Rigid Body dt ∂q̇ ∂q
Dynamics of

=( − τ1 ) ⋅ δq1 + ( − τ2 ) ⋅ δq2
an Open-chain d ∂L ∂L d ∂L ∂L
Manipulator − −
Newton-Euler
dt ∂q̇1 ∂q1 dt ∂q̇2 ∂q2

=( − τ1 ) ⋅ δq1 + ( − τ2 ) ⋅ (−A−1
2 A1 )δq1
Equations
d ∂L ∂L d ∂L ∂L
Coordinate- − −
invariant dt ∂q̇1 ∂q1 dt ∂q̇2 ∂q2
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


50

Chapter 4
Robot As δq1 ∈ Rn−k is free,
Dynamics and
Control

( − τ1 ) − AT1 A−T
2 ( − τ2 ) = 0
Lagrangian d ∂L ∂L d ∂L ∂L
Equations − −
Inertial
dt ∂q̇1 ∂q1 dt ∂q̇2 ∂q2
Properties of
Rigid Body

Dynamics of Lagrange-d’Alembert equation


an Open-chain
Manipulator

Newton-Euler
Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


51

◇ Example: Dynamics of a rolling disk


Chapter 4


Robot
Pfaffian constraint:

⎪ ẋ − ρ cos θ̇ ϕ̇ = 0
Dynamics and


Control ϕ



Lagrangian

⎩ ẏ − ρ sin θ ϕ̇ = 0
θ
Equations (x, y)
Inertial
Properties of

⇒ A(q)q̇ = [ 0 1 0 −ρ sin θ ] q̇ = 0
Rigid Body 1 0 0 −ρ cos θ
Dynamics of

L(q, q̇) = 21 m(ẋ2 + ẏ2 ) + 21 I1 θ̈ 2 + 21 I2 ϕ̇2


an Open-chain
Manipulator

Lagrange-d’Alembert equation:
⎡ m ⎤ ⎡ 0 ⎤
Newton-Euler

⎛⎢ ⎥ ⎢ 0 ⎥⎞
Equations

⎢ m ⎥ q̈ − ⎢ ⎥
⎢ ⎥ ⎢ τθ ⎥⎠ ⋅ δq = 0
Coordinate-

⎝⎢ ⎥ ⎢ ⎥
I1
⎣ 2 ⎦ ⎣ ϕ ⎦
invariant
algorithms for
robot
I τ
dynamics where A(q) ⋅ δq = 0.
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


52
As
δx = ρ cos θ ⋅ δϕ
Chapter 4 {
Robot δy = ρ sin θ ⋅ δϕ
Dynamics and
Control the equation of motion is:
Lagrangian
([ mρc0
θ mρsθ
0 ] [ ẍ ] + [ I1 0 ] [ θ̈ ] − [ τθ ]) ⋅ [ δθ ] = 0
ÿ 0 I2 τϕ δϕ
Equations
ϕ̈
⇒ [ mρc 0 ] [ ẍ ] + [ I1 0 ] [ θ̈ ] = [ τθ ]
Inertial
Properties of 0
Rigid Body θ mρsθ ÿ 0 I2 ϕ̈ τϕ


As

Dynamics of

⎪ ẍ = ρ cos θ ⋅ ϕ̈ − ρ sin θ ⋅ θ̇ ϕ̇
an Open-chain


Manipulator

⎪ ÿ = ρ sin θ ⋅ ϕ̈ + ρ cos θ ⋅ θ̇ ϕ̇


Newton-Euler
Equations

⇒ [ 01 I + mρ2 ] [ θ̈ ] = [ τϕθ ]
Coordinate- I 0 τ
invariant
2 ϕ̈
Solve for (θ(t), ϕ(t)), and then solve for (x(t), y(t)) from:
algorithms for


robot


⎪ ẋ = ρ cos θ ⋅ ϕ̇
dynamics


Lagrange’s


⇐ 1st order differential equation

Equations with

⎩ ẏ = ρ sin θ ⋅ ϕ̇
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


53

◻ Nature of nonholonomic constraints:


Consider q = (r, s) ∈ R2 × R with Pfaffian constraint,
Chapter 4

ṡ + aT (r)ṙ = 0, a(r) ∈ R2
Robot
Dynamics and
Control

Lagrangian Lagrangian
Equations
L = L(r, ṙ, ṡ)
Inertial
Properties of and constrained Lagrangian
Lc (r, ṙ) = L(r, ṙ, −aT (r)ṙ)
Rigid Body

Dynamics of
an Open-chain ⇒ Lagrange’s equation:
Manipulator
d ∂Lc ∂Lc
Newton-Euler − = 0, i = 1, 2
Equations dt ∂ṙi ∂ri
⎛ ∂L ∂L ∂aj ⎞
⇒ ( − ai (r) ) −
d ∂L ∂L

Coordinate-
ṙj = 0
⎝ ∂ri ∂ṡ j ∂ri ⎠
invariant −
algorithms for
robot
dt ∂ṙi ∂ṡ
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


54

∂L ⎛ ∂aj ⎞
⇒( ) − ai (r) ( − )= ȧi (r) − ∑ ṙj (∗)
d ∂L ∂L d ∂L ∂L
∂ṡ ⎝ j ∂ri ⎠
Chapter 4 −
Robot dt ∂ṙi ∂ri dt ∂ṡ ∂s
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Dynamics and
Control

Lagrangian
≠0
Equations If the constraint is holonomic, i.e.
Inertial

ai (r) =
Properties of ∂h
Rigid Body for some h ∶ E ↦ R
Dynamics of ∂ri
an Open-chain
Manipulator
then RHS (right hand side) of (∗) equals
Newton-Euler

∂L ⎛ ∂2 h ⎞
Equations
∂2 h
∑ ṙj − ∑
Coordinate-

∂ṡ ⎝ j ∂ri ∂rj j ∂rj ∂ri ⎠


invariant ṙj = 0
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


55
◻ Metric, duality and orthogonality on Tq E:
xn
Tq Q
Chapter 4
Robot ⋯
Dynamics and
x1 q 1
Control K = q̇T M(q)q̇
Lagrangian
2
Q = h−1 (0)
Equations 1
= ≪ q̇, q̇ ≫M
Inertial
Properties of
2
Rigid Body

Tq Q⊥ = {V1 ∈ Tq E∣ ≪ V1 , V2 ≫M = V1T MV2 = 0, ∀V2 ∈ Tq Q}


Dynamics of
an Open-chain

Tq∗ Q⊥ = {f ∈ Tq∗ E∣⟨f , V⟩ = 0, ∀V ∈ Tq Q} ∶ constraint forces


Manipulator

Newton-Euler

Tq E = Tq Q ⊕ Tq Q⊥ , Tq∗ E = Tq∗ Q ⊕ Tq∗ Q⊥


Equations

Coordinate-
invariant
algorithms for Definition:
M b ∶ Tq E ↦ Tq∗ E, ⟨M ♭ V1 , V2 ⟩ = V1T MV2 =≪ V1 , V2 ≫M
robot
dynamics

−1
M ♯ ∶ Tq∗ E ↦ Tq E, M ♯ = M ♭
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


56

Chapter 4
Robot
Dynamics and Reciprocal
Control

Lagrangian
Equations

Inertial
Properties of
Rigid Body
Property 4:
and dqi , i = 1, . . . , n of Tq E and Tq∗ E
Dynamics of

an Open-chain Under the basis ∂qi
respectively, the matrix representation of M ♭ and M ♯ is M and
Manipulator

Newton-Euler

M −1 respectively.
Equations

Coordinate-
invariant
algorithms for Property 5:
M ♯ (Tq∗ Q) = Tq Q
robot
dynamics

Lagrange’s
Equations with M ♯ (Tq∗ Q⊥ ) = Tq Q⊥
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


57

Given
Chapter 4 h ∶E ↦ Rk , m = n − k
Robot

h∗ ≜ Tq h ∶Tq E ↦ Th(q) Rk
Dynamics and
Control

Lagrangian
Equations h∗ ≜ Tq∗ h ∶Th(q)

Rk ↦ Tq∗ E
Inertial
Properties of we have
Rigid Body Property 6:
Dynamics of ker h∗ = Tq Q, h∗ (Tq Q⊥ ) = Th(q) Rk , h∗ (Th(q)

Rk ) = Tq∗ Q⊥
h∗
an Open-chain
Manipulator

Newton-Euler Tq∗ E Tq∗ Rk


Equations

M2♯ M2♯ = h∗ ○ M ♯ ○ h∗
Coordinate-
invariant
algorithms for M♯
robot
dynamics

Lagrange’s Tq E Th(q) Rk
Equations with
Constraints
h∗
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


58
Lemma 1: The map (I − Pω ) ∶ Tq∗ E ↦ Tq∗ Q⊥ given by
Chapter 4 (I − Pω ) = h∗ ○ M2♭ ○ h∗ ○ M ♯
Robot
Dynamics and is a well-defined projection map, with the property:
(I − Pω )f1 = 0, ∀f1 ∈ Tq∗ Q
Control

(I − Pω )f2 = f2 , ∀f2 ∈ Tq∗ Q⊥


Lagrangian
Equations

Inertial
Properties of Proof :
Given f1 ∈ Tq∗ Q, M ♯ (f1 ) ∈ Tq Q = ker h∗ , then (I − Pω )(f1 ) = 0. For
Rigid Body

f2 ∈ Tq∗ Q⊥ , ∃λ ∈ Rn−m s.t. f2 = h∗ λ, and


Dynamics of
an Open-chain

(I − Pω )f2 = h∗ M2♭ h∗ M ♯ h∗ λ
Manipulator

Newton-Euler

= h∗ λ = f2
Equations

⇒ Pω ∶ Tq∗ E ↦ Tq∗ Q
Coordinate-
invariant
algorithms for
robot is a well-defined projection map. Similarly,
PT ∶ Tq E ↦ Tq Q, PT = I − M ♯ h∗ M2b h∗
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


59

and (I − PT ) ∶ Tq E ↦ Tq Q⊥ are projection maps.


Chapter 4
Robot Lemma 2:
Dynamics and Pω M = MPT
Pω h∗ = h∗ PT = 0
Control

Lagrangian
Equations

Inertial PT = PTω
Properties of
Rigid Body

Dynamics of
an Open-chain
For nonholonomic constraints:
Manipulator h∗ ← A(q)
Newton-Euler
Equations h∗ ← A∗ (q)
Coordinate- Tq Q ← ∆ q
Tq∗ Q⊥ ← span{ai (q), i = 1, . . . , k}
invariant
algorithms for
robot
dynamics
application in hybrid velocity/force control.
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


60
◻ Lagrange’s equations of motion:
Chapter 4 M(q)q̈ + C(q, q̇) + N + AT (q)λ = τ
λ = (AM −1 AT )−1 (AM −1 (τ − C − N) + Ȧq̇)
Robot
Dynamics and
Control

Lagrangian
Equations
M(q)q̈ + AT (AM −1 AT )Ȧq̇ + Pω C + Pω N = Pω τ
Inertial Pω = I − AT (AM −1 AT )−1 AM −1
Properties of
Rigid Body

Dynamics of C̃ = Pω C
an Open-chain
Manipulator
Ñ = Pω N
Newton-Euler
Equations τ̃ = Pω τ
M θ̈ + AT (AM −1 AT )−1 Ȧθ̇ = Pω M θ̈ ≜ M̃ θ̈ ∶ intertia forces in Tq∗ Q
Coordinate-
invariant

Definition: Dynamics in Tq∗ Q


algorithms for
robot
dynamics

Lagrange’s M̃ θ̈ + C̃ + Ñ = τ̃
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


61
(I − Pω )(Mq̈ + C + N) = (I − Pω )τ + AT τ
Let
Chapter 4
Robot
PT = I − M −1 AT (AM −1 AT )−1 A
Dynamics and
Control
and
Pω M = MPT
∗ ⊥
Lagrangian
Equations
then the dynamics in T Q :
Inertial

M(I − PT )(q̈ + M −1 C) = (I − Pω )(τ − N) + AT λ


Properties of
Rigid Body

Dynamics of
an Open-chain ◻ Geometric Interpretation:
∇↔M
Manipulator

Mq̈ + C + N = τ + AT λ ⇔ M∇q̇ q̇ = τ − N + AT λ
Newton-Euler
Equations

Coordinate-
invariant ∇˜ ↔ induced metric on Tq Q
S ∶TQ ⊗ TQ ÐÐÐÐÐÐ→ N(Q) :2nd fundamental form
algorithms for
robot
dynamics

Lagrange’s ↑ ↑
Equations with
Constraints tangent vector field normal vector field
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


62

Chapter 4
Robot
∇X Y = ∇ ˜ X Y + S(X, Y)
−1
M (I − PT )(q̈ + M C) = (I − Pω )(τ − N) + AT λ
Dynamics and
Control

Lagrangian ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Equations S(q̇,q̇)
Inertial
Properties of
MS(q̇, q̇): centrifugal force due to curvature of Q in E
Rigid Body



Dynamics of



an Open-chain
˜ q̇ q̇ = τ̃ − Ñ
M∇
⎬ for hybrid control design
Manipulator

T ⎪
MS(q̇, q̇) = (I − Pω )(τ − N) + A λ ⎪

Newton-Euler


Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


63
◇ Example: Dynamics of a Spherical Pendulum

K = m(ẋ2 + ẏ2 + ż2 )


Chapter 4 1
Robot
Dynamics and
Control
2
1
Lagrangian = q̇T Mq̇
Equations 2
Inertial
Properties of
q = (x, y, z)T , M = mI
Rigid Body

Dynamics of
h ∶qT q − r2 = 0
an Open-chain

A = (x, y, z), M2♯ = AM ♯ AT = r2 /m


Manipulator

⎡ y2 + z 2 ⎤
Newton-Euler

1 ⎢ ⎥
Equations

Pω = 2 ⎢ ⎥
−xy −xz
⎢ x2 + z 2 ⎥
Coordinate-

r ⎢ −zx ⎥
−yz −yz
⎣ ⎦
invariant
algorithms for −zy x2 + y2
⎡ x2 xy xz ⎤
robot

1 ⎢ ⎥
dynamics

I − Pω = 2 ⎢ yx y2 yz ⎥
r ⎢ zx zy z2 ⎥

Lagrange’s


⎣ ⎦
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


64

PT = PTω = Pω
Chapter 4
(µ, ν): Spherical coordinates
q = (r cos µ cos ν, r cos µ sin ν, r sin µ)T
Robot
Dynamics and

˜ q̇ q̇ = PT (∇q̇ q̇)
Control

Lagrangian ∇
Equations
−r sin µ cos ν −r sin ν
Inertial
Properties of = [ −r sin µ sin ν r cos ν ] [ vv21 ]
Rigid Body r cos µ 0
Dynamics of
an Open-chain
S(q̇, q̇) = (I − PT )(∇q̇ q̇)
Manipulator r cos µ cos ν
Newton-Euler = (− µ̇2 − cos2 µ ν̇ 2 ) [ r cos µ sin ν ]
Equations r sin µ
Coordinate- where
invariant
algorithms for v1 = µ̈ + sin µ cos µ ν̇ 2
robot
dynamics v2 = cos µ ν̈ − 2 sin µ µ̇ν̇
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


65

Chapter 4
◻ Control Algorithm:
Robot
Dynamics and
Control
1 holonomic constraints:
Lagrangian
q̃ ∶ coordinates of Q
Equations
q = ψ(q̃) ⇒ q̇ = J ⋅ q̃˙
Inertial

τ = MJ(q̃¨ d − Kv ˙ẽ − Kp ẽ) + C1 + N + AT (−λd + KI ∫ (λ − λ ))


Properties of
Rigid Body d
Dynamics of
an Open-chain 2 nonholonomic constraints:
Let J(q) ∈ Rn×m be s.t. AJ = 0. Write q̇ = J ⋅ u for some u
Manipulator

τ = MJ(u̇d − Kp (u − ud )) + M J̇u + C + N + AT (−λd + KI ∫ (λ − λ ))


Newton-Euler
Equations
d
Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


66
◇ Example: 6-DoF manipulator on a sphere with
frictionless point contact
Chapter 4
Robot
Dynamics and
Control
Contact constraint:

vz = 0 ⇔ [0 0 1 0 0 0]Adg −1 Vof = 0
Lagrangian
Equations
fl f

⇒ Holonomic constraint:
Inertial
Properties of

η = (αoT , αfT , ψ): Parametrization of Q


Rigid Body

Dynamics of
an Open-chain Pω = diag(1, 1, 0, 1, 1, 1)
Manipulator

Newton-Euler Newton-Euler Equations of motion:


M V̇of − adTVof MVof = Fm + G + AT λ
Equations

⎡ Rψ Mo 0 ⎤
Coordinate-

⎢ −Mf ⎥⎡ ⎤
⎢ 0 ⎥ ⎢ α̇o ⎥
invariant

Vof = ⎢ Rψ Ro Ko Mo −Ro Kf Mf 0 ⎥ ⎢ α̇f ⎥ ≜ J η̇


algorithms for
0 0
⎢ ⎥ ⎢ ψ̇ ⎥
robot

⎢ −To Mo 1 ⎥ ⎣ ⎦
dynamics

Lagrange’s ⎣ −Tf Mf ⎦
MJ η̈ + C1 = Fm + G + AT λ
Equations with
Constraints ∗
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


67
Pω (∗) ∶
Chapter 4
M̃ η̈ + C̃1 = B1 Fm + B1 G
Robot
Dynamics and −ϕ3 − λ = b2 Fm + b2 G
F̂ = [f1 f2 f4 f5 f6 ]T = M̃(η̈d − Kv ˙ẽ − Kp ẽ) + C̃1 − B1 G
Control

Lagrangian

∫ (λ − λ ) − b G
Equations

Inertial f3 = −ϕ3 − λd + KI d 2
Properties of

τ = JsT Fm
Rigid Body

Dynamics of
an Open-chain
Manipulator

Newton-Euler
Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


68
◇ Example: 6-DoF manipulator rolling on a sphere
⎡ 1 0 0 0 0 0 ⎤
⎢ 0 1 0 0 0 0 ⎥
⎢ ⎥
⎢ 0 0 1 0 0 0 ⎥ Vlo lf = 0
Chapter 4

⎢ 0 0 0 0 0 1 ⎥
⎣ ⎦
Robot
Dynamics and

´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Control

Lagrangian
Equations
Ȧ1 (q)
T
Inertial fc = A1 λ, λ ∈ R4
Properties of
ωx
Rigid Body
[ ωy ] = −R0 (Kf + Rψ Ko Rψ )Mf α̇f
Dynamics of
an Open-chain
Manipulator Vof = Adgflf ⋅ Vlo lf
⎡ ⎤
⎢ 0 ⎥
Newton-Euler

⎢ ⎥
⎢ ⎥ α̇f
Equations
0
⎢ 0 ⎥
⎢ −Ro (Kf + Rψ Ko Rψ )Mf ⎥
Coordinate- = Adgflf
⎢ ⎥
invariant

⎣ ⎦
algorithms for
robot o
≜ Jf α̇f
dynamics

span{Jf }: Not involutive


Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


69

MJf α̈f + (M J̇f α̇f − adTJf α̇f MJf α̇f ) = Fm + G + AT λ


Chapter 4
Robot Fm = MJf (α̈fd − Kp (α̇f − α̇fd )) + (M J̇f α̇f − adTJf α̇f MJf α̇f )
Dynamics and

+ AT (−λd + ∫ (λ − λ )) − G
Control

Lagrangian
d
Equations

Inertial
Properties of ◇ Example: Redundant parallel manipulator
θ = (θ 1 , . . . , θ 6 ) ∈ E
Rigid Body

Dynamics of

θ a = (θ 1 , θ 3 , θ 5 )
an Open-chain
Manipulator

Newton-Euler
θ p = (θ 2 , θ 4 , θ 6 )
⎡ ⎤
Equations

⎢ xa + lc1 + lc12 − xb − lc3 − lc34 ⎥


⎢ ⎥
Coordinate-

H(θ) = ⎢ ⎥=0
ya + ls1 + ls12 − yb − ls3 − ls34
⎢ ⎥
invariant

⎢ ⎥
algorithms for
xa + lc1 + lc12 − xc − lc5 − lc56
⎣ ⎦
robot
dynamics ya + ls1 + ls12 − yc − ls5 − ls56
Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


70
Mi (θ) ∈ R2×2 : ith chain
Chapter 4 M(θ) = diag(M1 (θ), . . . , M3 (θ))
Robot
Dynamics and M(θ)θ̈ + C + N = τ + AT λ
Control
If all joints are actuated,
Lagrangian
Equations
Position control of end-effector
+
Inertial internal grasping force
Properties of
As τ2 , τ4 , τ6 = 0,
θ̃ ∈ R2 ∶ local parametrization of Q = H −1 (0)
Rigid Body

Dynamics of
an Open-chain
Manipulator θ = ψ(θ̃) ∶embedding of Q in E
Newton-Euler
Equations
θ̇ = J θ̃˙
Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints
Chapter 4 Robot Dynamics and Control

4.6 Lagrange’s Equations with Constraints


71

Given Pω ∶ Tθ∗ E ↦ Tθ∗ Q, the dynamics in Tθ∗ Q is given by:


Chapter 4

Pω MJ θ̃¨ + Pω (C1 + N) = Pω τ
Robot
Dynamics and

τ̃ = (τ1 , τ3 , τ5 )
Control

Lagrangian

P̃ω = (P1 , P3 , P5 )
Equations

Inertial
Properties of
Rigid Body τ̂ = P̂ω τ̃ = Pω τ ∈ R6
τ̂ = Pω MJ(θ̃¨d − Kv ˙ẽ − Kp ẽ) + Pω (C1 + N)
Dynamics of
an Open-chain
Manipulator

Newton-Euler
Equations

Coordinate-
invariant
algorithms for
robot
dynamics

Lagrange’s
Equations with
Constraints

You might also like