Lecture 2 - Dynamics - Part 3 and 4

Dynamics of Orbital Systems
(part III ad IV)

Summer Semester 2018
TU Munich May 2018
[email protected]
Contents
Dynamics is the study of the causes of motion and changes in motion.

In other words the study of forces and why objects are in motion.
We are interested in describing the causes and changes of motion of a

point mass, of a single rigid body and of a multi-body system (e.g., a
robot) in Earth orbit.
In the frame of Robotics, we introduce the forward and inverse

dynamics problems and the Generalized Jacobian for a free-floating
robot.
We then describe the uses of the kinematic and dynamic models of a

space system for engineering purposes: simulation, open-loop control,
closed-loop control.
Folie 2
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2016 | Dynamics
Equations of motion of a multi-body system
and robot dynamics
Folie 3
Geometry of a free-flying robot as a multibody system
Folie 4
Differential kinematics of a robot with a moving base
Recall from Lesson 1 that the differential kinematics of a robot with revolute joints
are given by 𝑛𝑛 𝑗𝑗
𝑣𝑣 𝑒𝑒 = ∑𝑗𝑗=1 𝝎𝝎
� 𝑙𝑙 𝑗𝑗
with ω𝑖𝑖 = ∑𝑖𝑖𝑗𝑗=1 ω𝑗𝑗(𝑗𝑗−1) = ∑𝑖𝑖𝑗𝑗=1 𝜃𝜃̇ 𝑗𝑗 𝑢𝑢 𝑗𝑗
and ω𝑒𝑒 = ∑𝑛𝑛𝑗𝑗=1 𝜃𝜃̇ 𝑗𝑗 𝑢𝑢 𝑗𝑗
For a robot with a moving base, the motion of the base must also be accounted
for. The expression which relates v , ω , v , ω and θ is
e e 0 0
𝑗𝑗
𝑣𝑣 𝑒𝑒 = 𝑣𝑣 0 + ∑𝑛𝑛𝑗𝑗=1 𝝎𝝎
� 𝑙𝑙 𝑗𝑗
with ω𝑖𝑖 = ω0 + ∑𝑖𝑖𝑗𝑗=1 ω𝑗𝑗(𝑗𝑗−1) = ω0 + ∑𝑖𝑖𝑗𝑗=1 𝜃𝜃̇ 𝑗𝑗 𝑢𝑢 𝑗𝑗
and ω𝑒𝑒 = ω0 + ∑𝑛𝑛𝑗𝑗=1 𝜃𝜃̇ 𝑗𝑗 𝑢𝑢 𝑗𝑗
Folie 5
Differential kinematics in matrix form
This can be expressed in matrix form as:
x e = J b x 0 + Jθ
with x = [v
e e
ω e ]T , x 0 = [v 0 ω 0 ]T . The expression for the base Jacobian Jb
(6x6) is
I 3 −~
r e0 
Jb =  
0 I3 
n
where I3 is the identity matrix (3x3) an r
e0
= ∑ l i , i.e. the vector from the base
i =0
centre of mass to the end-effector.
Note: see a Jacobian as any matrix which linearly multiplies two velocity vectors
of a mechanical system. For example, given the velocity of the base
x 0, the velocity of the end-effector which results from it is J b x 0.
Folie 6
General equations of motion of a free-flying robot
Fe2 Fe
Fe3
We can write the equations of motion for a
free-flying robot with n joints as {Oe3,ee3}
𝐉𝐉bT e
𝐅𝐅 0
𝐌𝐌𝐲𝐲̇ II + 𝐂𝐂𝐲𝐲II = + T 𝐅𝐅
𝛕𝛕 𝐉𝐉
θi
[ ]
T
where y II = v ω θ ... θ ,
0 1 0 n
of dimensions (6+n)x1. F0
yI=[r0 φ0 θ]
Recall the derivation of the
{OI,eI}
equations for a two-body system in
Lesson 2 Part I and II, S. 35-41.
Modelling of a general multibody
open-tree system with n joints
Folie 7
To derive the second term on the RHS, recall the result that (see slide 41, Part I):
𝐅𝐅 = 𝛗𝛗𝑇𝑇 𝐅𝐅� where x II = φ y II
In this case, 𝐅𝐅� = 𝐅𝐅 𝒆𝒆 , to say that the external forces/torques of interest are those
acting on the last link.
We then derive the expression for φ for the last link, i.e. for the end-effector body:
letting then  v 0

v 
e
   ve   v0  
x II =  e , y II = ω0  and given that  e  = J b  0  + Jθ
ω   θ  ω  ω 
 
it follows that
J b 
φ= 
 J 
Folie 8
𝐅𝐅 0 𝐉𝐉bT e
𝐌𝐌𝐲𝐲̇ II + 𝐂𝐂𝐲𝐲II = + T 𝐅𝐅 Fe2 Fe1
𝛕𝛕 𝐉𝐉 Fe3
Notable properties: {Oe3,ee3}

Matrix 𝐌𝐌: the inertia matrix
• symmetric, positive definite
• configuration dependent, θi
i.e. 𝐌𝐌 = 𝐌𝐌(𝛟𝛟0 , 𝛉𝛉) 0
F
• Dimensions: (6+n)x(6+n). yI=[r0 φ0 θ]
Matrix 𝐂𝐂:
{OI,eI}
• includes the quadratic velocity
terms (Coriolis and centrifugal), Modelling of a general multibody
0 ̇
i.e. 𝐂𝐂 = 𝐂𝐂(𝛚𝛚 , 𝛉𝛉) open-tree system with n joints
• Dimensions: (6+n)x(6+n).
Column matrix 𝐅𝐅0:
• includes all external forces and torques on the base body.
Folie 9
𝐅𝐅 0 𝐉𝐉bT e
𝐌𝐌𝐲𝐲̇ II + 𝐂𝐂𝐲𝐲II = + T 𝐅𝐅 Fe2 Fe1
𝛕𝛕 𝐉𝐉 Fe3
Column matrix 𝛕𝛕 : {Oe3,ee3}

• includes the internal applied
torques on the joints.
Column matrix 𝐅𝐅e: θi
• includes all external forces and 0
F
torques on the end-effector yI=[r0 φ0 θ]
(only one considered here).
{O I,eI}
Matrix 𝐉𝐉b:
• the Jacobian matrix between the Modelling of a general multibody
base velocity and the
open-tree system with n joints
end-effector velocity.
Matrix 𝐉𝐉:
• the Jacobian matrix between the joint velocities
and the end-effector velocity (see Lesson 1).
Folie 10
Direct and inverse dynamics
The direct and the inverse dynamics problems have to do with the relation
between accelerations and forces/torques:
in the first, the joint torques are given and we want to compute the joint
accelerations which derive from them
in the second, the accelerations of the joints are given and we want to
compute the torques which produce them
In more detail, the direct dynamics problem consists of determining, for t > t0, y II
and then y II (t ) and y I (t ) through integration, resulting from the given base
body forces F0(t), joint torques τ , end-effector forces Fe and initial conditions
y I (0), y II (0) : T
0 𝐉𝐉
𝐲𝐲̇ II = 𝐌𝐌−𝟏𝟏 𝐅𝐅 + bT 𝐅𝐅 e − 𝐂𝐂𝐲𝐲II y II = y II dt ∫
𝛕𝛕 𝐉𝐉
The inverse dynamics problem consists of determining the base body forces F0(t)
and the joint torques τ(t) which are needed to generate the motion specified by
y I (t ), y II (t ), y II (t ), once the end-effector forces Fe are known:
T
0
𝐅𝐅 = 𝐌𝐌𝐲𝐲̇ + 𝐂𝐂𝐲𝐲 − b 𝐅𝐅 e 𝐉𝐉
II II T
𝛕𝛕 𝐉𝐉
Folie 11
Uses of the kinematic and dynamic models
Solving the direct dynamics problem is useful for system simulation. Given the
input control actions and external disturbances or contact forces, we integrate
the system dynamical and kinematical equations of motion to arrive at a
description of its motion on position and velocity level.
Solving the inverse dynamics problem is useful for control algorithm

implementation (open or closed loop). Once the trajectory of the system DOFs is
given, in terms of the positions, velocities and accelerations, we compute the
necessary actuation forces and torques to obtain the desired motion.
Both kinematical and dynamical models are meant to represent the physical
system as closely as possible, to minimize the modelling errors. Often though,
due to complexity or computational burden, simplifying modelling assumptions
are or have to be made (e.g. rigid body assumption).
Folie 12
Integration of the differential equations (ODEs)
Solution of an initial value problem:
Kinematics:
d𝛉𝛉  yI  t  y I 
𝑟𝑟̇ 0 = 𝑣𝑣 0 , 𝜙𝜙̇ = 𝓒𝓒 𝜙𝜙 0 𝜔𝜔0 , = 𝛉𝛉̇ y  ∫ y (y , y ) dt
=
dt  II  0  II I II 
Dynamics: 𝐅𝐅 0 𝐉𝐉bT e
𝐌𝐌𝐲𝐲̇ II + 𝐂𝐂𝐲𝐲II = + T 𝐅𝐅 given F0(t), τ(t), Fe(t)
𝛕𝛕 𝐉𝐉
Initial conditions:
r 0 (0) = r00 , v 0 (0) = v00 , φ 0 (0) = φ00 , ω 0 (0) = ω00 , θ(0) = θ 0 , θ (0) = θ 0
Integration methods:
Runge-Kutta order 5 (define accuracy, may have variable step size, etc.)
Euler (order 1) is generally too inaccurate
Other
Example: to guarantee the conservation of momentum for the free-floating
dynamics case (see later), the integrator accuracy must be in the range 10-8.
Folie 13
Example: 7 DOF robot dynamics
Parameter description [kg, kg m2]:
mass m i
inertia i I i
SpaceDyn dynamic model

7 DOF Manipulator
Folie 14
Dynamical properties
of a free-floating robot
Folie 15
TECSAS (2001-2005) -Technology Satellite for
Demonstration and Verification of Space Systems
Target spinning axis

(ω = 4 deg/sec)
Target satellite
mTARGET ≈ mCHASER
Manipulator
SARAH hand
Handle
(7 D.O.F.)
Chaser satellite
(thruster controlled)
Folie 16
TECSAS - Grasping Task
• Very low Servicer-to-Manipulator

mass ratio
• Servicer and Client with
flexible appendages therefore
high risk of collisions
Folie 17
DEOS Deutsche Orbitale Servicing Mission
System description:
• Servicer satellite
• 7 DOF Manipulator
• Client satellite (cylinder with no appendages)
Operational conditions:
• Servicer attitude control inactive
• Client motion constrained to limited
range of free tumbling states
Robot control modes for the Grasping Task:

• Tele-presence
• Semi-autonomy
Example of Grasping Task
Folie 18
Definition of a Free-floating and a Free-flying robot
A free-floating robot is one for which the base body is in free flight. Furthermore,
all external forces are zero. This includes environmental forces, such as orbital
disturbances, as well as control forces applied to the base body, deriving from
actuators such as thrusters or reaction wheels.
As a result, as we will see, the momentum of the robot system is conserved and
a free-floating dynamics results. We will now analyse this dynamics.
A free-flying robot is one for which the base body is generally not attached to
ground, as opposed to a fixed-base robot. It may however, other than with a free-
floating robot, be subjected to control forces applied to the base body and to
other external forces.
Folie 19
Linear and Angular Momentum of a multibody system
We recall that the linear and angular momentum vectors of a multibody system
are defined as (see L.2-I-II, S.28-29):
n
P = ∑ m j v CM
j
j =0
CM H=∑
n
i =0
[CM i
I . ω + m (d × v
i i i i CM i
) ]
Note that both quantities are functions of the velocity variables v , ω , θ .
0 0
In fact, recalling the differential kinematics expressions for the end-effector (S. 5),
we can similarly write for the velocity of the centre of mass of the i th body:
𝑖𝑖
𝑣𝑣 𝐶𝐶𝑀𝑀 = 𝑣𝑣 0 + ∑𝑖𝑖−1 � 𝑗𝑗 𝑙𝑙 𝑗𝑗 + 𝜔𝜔
𝑗𝑗=1 𝜔𝜔 � 𝑖𝑖 𝑐𝑐 𝑖𝑖
with 𝜔𝜔𝑗𝑗 = 𝜔𝜔0 + ∑𝑖𝑖−1
𝑘𝑘=1 𝜃𝜃̇ 𝑘𝑘 𝑢𝑢 𝑘𝑘 and 𝜔𝜔𝑖𝑖 = 𝜔𝜔 + ∑𝑖𝑖
0
̇ 𝑘𝑘 𝑘𝑘
𝑘𝑘=1 𝜃𝜃 𝑢𝑢
These are linear functions in the velocities which allow a simple matrix formulation.
Folie 20
Generalized Jacobian (1/3)
The linear and angular momentum can in fact be written in matrix form, as
P
L =   = F0 x 0 + G 0 θ (1)
H 
where F0 and G0 are the appropriate corresponding matrices and L is a vector
array containing vectors P and H.
F0 and G0 can be shown to be related to the inertia matrix of the multibody
system M. In fact, writing the equations of motion for the free-floating case, for
which no external forces and torques act :
Mb (6x6) – inertia matrix
Partitioning of M  M b M bm  x   C b  0 
0
of base body
M T + =
    C   τ 
m  θ 
and C matrices Mm (nxn) – inertia matrix
 bm M  m   of robot arm
𝐂𝐂b Mbm (6xn) - dynamic
where = 𝐂𝐂𝐲𝐲II , it follows that: coupling term
𝐂𝐂m
M bx0 + M bmθ + C b = 0 (2)
(6x6) (6x1) + (6xn)(nx1) + (6x1) = 0
Folie 21
Differentiating (1) in time we get
dL  0  θ + G θ = 0
= F0 x + F0x0 + G 0 0
(3)
dt
where the derivative of L is zero, given that there are no external forces acting.
Equating coefficients between (3) and (2), it follows that
F0 = M b G 0 = M bm and  θ = C
F 0 x 0 + G 0 b
This is also equivalent to saying that equation (1) is solution (first integral) to the
equations of motion (2), i.e. for the case of zero external forces, for which the
linear and angular momentum of the system are constant in time.
Folie 22
From (1), the dependence of x e on x 0 can be eliminated. We first write from (1)
x 0 = −M -1 M θ + M -1 L
b bm
(4)
b
 = J b x + Jθ
Recalling again (S. 5) that: x
e 0
 0 with the expression in (4), we obtain

and substituting x
x e = J gθ + x e0
where
J g = J − J bM -1bM bm - generalized Jacobian, function of M b and M bm
x e0 = J bM -1bL
x e = J gθ
- differential kinematics for

If L = 0, then x
e0
=0 and a free-floating robot
The generalized Jacobian of a free-floating robot is the equivalent of the

Jacobian for a fixed-based robot. However, it depends on the inertial properties
of the multibody system, as well as on its kinematical properties.
Folie 23
Simulation Example – point-to-point maneuver
𝐱𝐱̇ 𝑒𝑒 (𝑡𝑡) 𝛉𝛉̇(𝑡𝑡)
t=0 t=tf
tf tf
𝐱𝐱 𝑒𝑒 = � 𝐉𝐉g 𝛉𝛉̇ 𝑑𝑑𝑑𝑑 or 𝛉𝛉 = � 𝐉𝐉g−1 𝐱𝐱̇ 𝑒𝑒 𝑑𝑑𝑑𝑑
0 0
(𝛉𝛉̇(t) given) (𝐱𝐱̇ 𝑒𝑒 (t) given)
Folie 24
Video – point-to-point maneuver
x e = J gθ
Folie 25
Dynamic singularities
The dynamic singularities correspond to configurations for which the generalized
Jacobian is rank-deficient and the end-effector again loses mobility in some
inertial direction.
They are affected by the dynamic coupling between the robot and its free-floating
base. In fact, the expression of the generalized Jacobian includes a term which is
function of the inertia and inertia coupling terms (see S.23). Note however, that
for a given system, the rank of the Jacobian is solely configuration dependent.
The generalized Jacobian of course also has a null-space, if the robot is

kinematically redundant, such that one can write
θ = J g⊥ v e + (I n − J g⊥ J g ) v 0
with an arbitrary vector v0 multiplying the null-space term (see Lesson 1), which
can be used to move away from singularities.
Folie 26
Linear Momentum of a Free-floating Robot
The momentum equations of a free-floating robot are again:
L = M b x 0 + M bm θ (5)
[ ]
T
with x = v ω . We can extract the first three equations from (5), relative to
0 0 0
the linear motion, and write them as
P=M ˆ θ
ˆ x 0 + M (6)
b bm
(3x1) = (3x6)(6x1) + (3xn)(nx1)
This acts like a motion constrain between 𝐱𝐱̇ 0 and 𝜽𝜽̇ (note that P is constant). It
can be integrated analytically to express the fact that the centre of mass of the
system is either stationary (𝐏𝐏 = 𝟎𝟎) or moves with a constant velocity (𝐏𝐏 ≠ 𝟎𝟎):
d  n j CM j  d
( )
n
∑m v j CM j
(t ) =  ∑ m r (t )  = Mr CM (t ) = P = const.
j =0 dt  j =0  dt
(see also L. 2-I-II, S. 9-10).
As a result, for any new configuration of the robot, the position of the base body
can be determined, given the center of mass position of the system.
Folie 27
Angular Momentum of a Free-floating Robot
We can now extract the last three equations from (5), relative to the angular
motion, and write them as
~ 0 ~ 
H = M x + M θ
b bm
(7)
(3x1) = (3x6)(6x1) + (3xn)(nx1)
This is also a motion constraint between x and θ (note that H is constant),
0
which however cannot be integrated analytically. It presents as such a non-

holonomic constraint.
A non-holonomic constraint is one of the form
g i (x, t ) = 0, i = 1,..., nc , nc ≤ n, x ∈ ℜ n (8)
and is non-integrable. It requires that the LHS of (8) be not an exact differential,
for which a function Q would exist, such that
dQ = g (x, t ).
Note that when the differential dQ is exact, its integral is
t - dependent only on the boundary conditions
∫ dQ = Q(t ) − Q(0)
0
- independent of the path followed inbetween
them
Folie 28
While such a property can be attributed to the conservation of linear
momentum, this does not hold for the angular momentum. Let us derive an
expression for 𝜔𝜔0 . Starting from Eq. (5), appropriately partitioned:
𝐏𝐏 � b𝑣𝑣 0 𝐌𝐌
𝐌𝐌 � b𝜔𝜔0 𝑣𝑣 0 � bm
𝐌𝐌
= 0 + � 𝛉𝛉̇ (9)
𝐇𝐇 � �
𝐌𝐌b𝑣𝑣 0 𝐌𝐌b𝜔𝜔0 𝜔𝜔 𝐌𝐌bm
we eliminate the dependence of 𝜔𝜔0 on 𝑣𝑣 0 , and assume P = H = 0 for
simplicity. We first extract the first row of (9) to get
� b𝑣𝑣 0 𝑣𝑣 0 + 𝐌𝐌
𝐌𝐌 � bm 𝛉𝛉̇ = 𝟎𝟎
� b𝜔𝜔0 𝜔𝜔0 + 𝐌𝐌
from which an expression for 𝑣𝑣 0 follow as
𝑣𝑣 0 = 𝐌𝐌 −1
� b𝑣𝑣 0 (−𝐌𝐌 � bm 𝛉𝛉̇)
� b𝜔𝜔0 𝜔𝜔0 − 𝐌𝐌 (10)
The second row of (9) is instead
� b𝑣𝑣 0 𝑣𝑣 0 + 𝐌𝐌
𝐌𝐌 � bm 𝛉𝛉̇ = 𝟎𝟎
� b𝜔𝜔0 𝜔𝜔0 + 𝐌𝐌 (11)
Substituting (10) in (11) and rearranging, we get
� b𝜔𝜔0 − 𝐌𝐌
𝐌𝐌 � b𝑣𝑣 0 𝐌𝐌
� b−1 �
𝐌𝐌 𝜔𝜔 �
0 = −(𝐌𝐌 − �
𝐌𝐌 �
0 𝐌𝐌
−1 � ̇
𝑣𝑣 0 b𝜔𝜔 0 bm b𝑣𝑣 b𝑣𝑣 0 𝐌𝐌bm )𝛉𝛉
Folie 29
or
� b# 𝜔𝜔0 = −𝐌𝐌
𝐌𝐌 � bm
#
𝛉𝛉̇
� # = (𝐌𝐌
where 𝐌𝐌 � 0 − 𝐌𝐌
� b𝑣𝑣 0 𝐌𝐌
� −10 𝐌𝐌 � bm
� 0 ) and 𝐌𝐌 # � bm − 𝐌𝐌
= 𝐌𝐌 � b𝑣𝑣 0 𝐌𝐌
� −10 𝐌𝐌
� bm , to
b b𝜔𝜔 b𝑣𝑣 b𝜔𝜔 b𝑣𝑣
obtain
−𝟏𝟏
𝜔𝜔0 � b#
= − 𝐌𝐌 � bm
𝐌𝐌 #
𝛉𝛉̇
Recalling that 𝜔𝜔 = 𝓒𝓒 𝛗𝛗 𝛗𝛗̇ (L.1, S. 21) and integrating in time we obtain

tf
−1
𝛗𝛗0 � b#
𝑡𝑡 = � 𝓒𝓒(𝛗𝛗0 ) 𝐌𝐌 � bm
𝐌𝐌 #
𝛉𝛉̇ 𝑡𝑡 𝑑𝑑𝑑𝑑
0
The resulting integral is path dependent (the integrand is not an exact

differential). This implies in practice that the final orientation of the base,
𝛗𝛗0 t f , is dependent on the path of the robot (i.e., of the integral), described
by 𝛉𝛉̇ 𝑡𝑡 , 0 ≤ 𝑡𝑡 ≤ t f .
Folie 30
Free-floating robot dynamics – point-to-point maneuver
Final base body

orientation
Initial configuration: 𝜃𝜃1 = 0, 𝜃𝜃2 = 0, 𝜃𝜃3 = 0 is path dependent
Path and final configuration 1: 𝜃𝜃1 = 𝜃𝜃11 , 𝜃𝜃2 = 𝜃𝜃12 , 𝜃𝜃3 = 0
Path and final configuration 2: 𝜃𝜃1 = 𝜃𝜃21 , 𝜃𝜃2 = 𝜃𝜃22 , 𝜃𝜃3 = 0
Folie 31
Free-floating robot dynamics – closed-loop maneuver
- ∆φ20 P
∆φ10
∆φ
If 𝐎𝐎𝐎𝐎1: Δ𝜙𝜙10
𝐎𝐎𝐎𝐎2: Δ𝜙𝜙20
and Δ𝜙𝜙10 ≠ Δ𝜙𝜙20
Initial configuration: 𝜃𝜃1 = 0, 𝜃𝜃2 = 0, 𝜃𝜃3 = 0
then
𝐎𝐎𝐎𝐎1 −𝐎𝐎𝐎𝐎2:
Final configuration: 𝜃𝜃1 = 0, 𝜃𝜃2 = 0, 𝜃𝜃3 = 0
Δ𝜙𝜙= Δ𝜙𝜙10 - Δ𝜙𝜙20 ≠ 0
Folie 32
Simulation Example – closed-loop maneuver
At the end of each closed-loop the robot returns to the same initial configuration.
However, the base body attitude changes.
Folie 33
Video – closed-loop maneuver
Note that the example on the left does not give rise to a base body attitude change.
The robot motion in this case is not a closed-loop motion but a “back and forth”
motion on a fixed path in joint space.
Folie 34
ETS-VII robot “attitude control” maneuvers (1998)
SIMPACK
simulation
model
Robot joint space
Satellite attitude
Measured
attitude
change
Folie 35
Free-floating and free-flying robot DOFs
Free-flying robot – base body is actuated:
- degrees of freedom = 6+n
Free-floating robot – no external forces:
Conservation of linear momentum (holonomic constraint):

- degrees of freedom = 3+n - described by (3+n)-D sub-manifold Qff
- independent velocities = 3+n
Conservation of angular momentum (nonholonomic constraint):

- degrees of freedom = 3+n
- independent velocities = n
Accessibility of sub-manifold Qff is preserved
Folie 36
Literature
Robertson, Schwertassek, “Dynamics of Multobody Systems”, 1988.

Sciavicco, Siciliano, “Modeling and Control of Robot Manipulators”, 1996
Edition.
Hughes, “Spacecraft Attitude Dynamics”, 2004.
Handbook of Robotics, Chapter 45, 2008.
Featherstone – “Robot Dynamics Algorithms”, 1987.
Xu, Kanade (Editors), “Space Robotics: Dynamics and Control”
Folie 37

Lecture 2 - Dynamics - Part 3 and 4

Uploaded by

Copyright:

Available Formats

Lecture 2 - Dynamics - Part 3 and 4

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture 2 - Dynamics - Part 3 and 4

Uploaded by

Copyright:

Available Formats

Dynamics of Orbital Systems

(part III ad IV)

TU Munich May 2018

Dynamics is the study of the causes of motion and changes in motion.

We are interested in describing the causes and changes of motion of a

In the frame of Robotics, we introduce the forward and inverse

We then describe the uses of the kinematic and dynamic models of a

Notable properties: {Oe3,ee3}

Column matrix 𝛕𝛕 : {Oe3,ee3}

Solving the inverse dynamics problem is useful for control algorithm

SpaceDyn dynamic model

Target spinning axis

• Very low Servicer-to-Manipulator

Robot control modes for the Grasping Task:

Differentiating (1) in time we get

 0 with the expression in (4), we obtain

The generalized Jacobian of a free-floating robot is the equivalent of the

The generalized Jacobian of course also has a null-space, if the robot is

the linear motion, and write them as

which however cannot be integrated analytically. It presents as such a non-

Recalling that 𝜔𝜔 = 𝓒𝓒 𝛗𝛗 𝛗𝛗̇ (L.1, S. 21) and integrating in time we obtain

The resulting integral is path dependent (the integrand is not an exact

Final base body

Path and final configuration 1: 𝜃𝜃1 = 𝜃𝜃11 , 𝜃𝜃2 = 𝜃𝜃12 , 𝜃𝜃3 = 0

Path and final configuration 2: 𝜃𝜃1 = 𝜃𝜃21 , 𝜃𝜃2 = 𝜃𝜃22 , 𝜃𝜃3 = 0

Free-floating robot – no external forces:

Conservation of linear momentum (holonomic constraint):

Conservation of angular momentum (nonholonomic constraint):

Accessibility of sub-manifold Qff is preserved

Robertson, Schwertassek, “Dynamics of Multobody Systems”, 1988.

You might also like