MTS-417 Introduction to

Robotics

Lecture 5
Velocities, Static Forces and
Jacobian

1
Outline
• Differentiation of a position vector

• Linear and Rotational Velocities of Rigid Bodies

• Motion of links of a robot: velocity ‘propagation’

• Jacobians

• Singularities

• Static Forces in Manipulators

• Jacobian in Force Domain

2
Time Varying Position and Orientation
Differentiation of a Position Vector
a. Notation for the derivative of the vector (Q)

Derivative of a point Q relative to frame {B}

• Velocity of a position vector >>>>>>>>>>> Linear velocity of a point in space
• If Q is not changing relative to {B}, then the velocity in {B} is zero, even if Q is changing wrt any other
frame
• Important to indicate the frame of differentiation
b. Velocity vector expressed in frame {A}
Differentiation done wrt frame {B}
Resulting vector expressed wrt frame {A}

Leading superscript can be removed by using R

Velocity of origin of frame {C} wrt universal frame

3
Time Varying Position and Orientation
Differentiation of a Position Vector
b. Angular Velocity Vector (𝛀)
• As v describes an attribute of a point, Ω describes an attributes of a body
• As frame is attached rigidly to the body, Ω define rotational motion of a frame
• 𝐴Ω describes the rotation of {B} wrt {A}
B
• Direction of 𝐴ΩB indicates the instantaneous axis of rotation of {B} wrt {A}
• Magnitude of 𝐴ΩB indicates speed of rotation

𝐶 𝐴
Ω B = Angular velocity of {B} wrt {A} expressed in {C}

𝜔𝐶 = 𝑈ΩC = Angular velocity of {C} wrt {U}

4
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
a. Linear Velocity
• Lets describe the motion of BQ wrt {A}
• Frame {B} is located relative to frame {A}, as described by 𝐴𝑃𝐵𝑂𝑅𝐺 and 𝐴
𝐵𝑅
• Assume 𝐴𝐵𝑅 is not changing wrt time.
BQ

5
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
b. Derivation of an orthonormal matrix
• If R is a proper orthonormal matrix, or rotation matrix of order n x n
𝑅𝑅𝑇 = 𝐼𝑛 = n x n Identity matrix
Differentiate
ሶ 𝑇 + 𝑅𝑅ሶ 𝑇 = 0 (n x n matrix)
𝑅𝑅
𝑅𝑅ሶ 𝑇 + (𝑅𝑅
ሶ 𝑇 )𝑇 = 0 3 x 3 (using (AB)T=BTAT)

Define ሶ 𝑇 = skew symmetric matrix (𝑆 + 𝑆 𝑇 = 0) and RT=R-1

S = 𝑅𝑅

Therefore 𝑆 = 𝑅ሶ 𝑅−1 ՜ 𝑅ሶ = 𝑆𝑅

6
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
c. Velocity of point due to rotating reference frame
Consider a fixed vector 𝐵𝑝 unchanging wrt fixed frame {B}. Its description in another frame {A}
is given as
𝐴𝑝 = 𝐵𝐴𝑅 𝐵𝑝

Differentiate 𝐴𝑝ሶ = 𝐵𝐴𝑅ሶ 𝐵𝑝 + 𝐵𝐴𝑅 𝐵𝑝ሶ

𝐴𝑝ሶ 𝐴 𝐴 𝐵
= 𝐵𝑆 𝐵𝑅 𝑝

= 𝐵𝐴𝑆 𝐴𝑝
𝐴𝑝ሶ

We have derived two important results 𝑝ሶ = 𝑆𝑝 and 𝑅ሶ = 𝑆𝑅

S is called as the angular velocity vector.

7
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
d. Skew symmetric matrix and vector-cross product
Elements to the skew symmetric matrix ‘S’

0 −Ω𝑧 Ω𝑦 Ω𝑥
𝑆 = Ω𝑧 0 −Ω𝑥 where 𝛀 = Ω𝑦
−Ω𝑦 Ω𝑥 0 Ω𝑧
𝑝𝑥 −Ω𝑧 𝑝𝑦 + Ω𝑦 𝑝𝑧
Lets take a vector 𝒑 = 𝑝𝑦 , then 𝑆𝒑 = Ω𝑧 𝑝𝑥 − Ω𝑥 𝑝𝑧 ,
𝑝𝑧 −Ω𝑦 𝑝𝑥 + Ω𝑥 𝑝𝑦

𝒊 𝒋 𝒌 Ω𝑦 𝑝𝑧 − Ω𝑧 𝑝𝑦
Ω × 𝒑 = Ω𝑥 Ω𝑦 Ω𝑧 = Ω𝑧 𝑝𝑥 − Ω𝑥 𝑝𝑧
𝑝𝑥 𝑝𝑦 𝑝𝑧 Ω𝑥 𝑝𝑦 − Ω𝑦 𝑝𝑥
The equations 𝐴 𝐴 𝐴
𝑃𝑉 = 𝐵𝑆 𝒑 velocity can be
written as 𝐴𝑉 = 𝐴Ω × 𝐴𝒑
𝑃 𝐵
8
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
e. Physical insight of the angular-velocity vector

There exists some vector ‘Ω’ such that 𝐴𝑃𝑉 = 𝐴 × 𝐴𝒑

𝐵Ω

𝑅ሶ = lim [𝑅 𝑡 + ∆𝑡 − 𝑅(𝑡)] /∆𝑡

∆𝑡՜0

𝑅 𝑡 + ∆𝑡 can be written as 𝑅 𝑡 + ∆𝑡 = 𝑅𝐾 ∆𝜃 𝑅 𝑡

(𝑅𝐾 ∆𝜃 − 𝐼3 )
𝑅ሶ = lim 𝑅(𝑡)
∆𝑡՜0 ∆𝑡

From equivalent axis rotation concept….

9
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
e. Physical insight of the angular-velocity vector

f. where 𝑐𝜃 = 𝑐𝑜𝑠𝜃 = 1
e. 𝑠𝜃 = 𝑠𝑖𝑛𝜃 = ∆𝜃
𝑣𝜃 = 1 − 𝑐𝑜𝑠𝜃 = 0

10
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
e. Physical insight of the angular-velocity vector
Dividing by ∆𝑡 and taking limit

Multiplying bothe sides by R-1

• The physical meaning of the angular-velocity vector Ω is that, at any instant, the change in
෡
orientation of a rotating frame can be viewed as a rotation about some axis 𝐾.

• This instantaneous axis of rotation, taken as a unit vector and then scaled by the speed of
ሶ yields the angular-velocity vector.
rotation about that axis (𝜃),
11
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
f. Rotational Velocity
Only Rotational Velocity ( 𝐵𝑉𝑄 = 0 if Q is viewed from {B})
• Fames with coincident origins (forever) with zero linear velocity
• The orientation of {B} is changing in time wrt {A} and denoted by 𝑨𝜴𝑩

𝑑 𝐴 𝐵
𝐴𝑄 = 𝐴 𝐵
𝐵𝑅 𝑄 Differentiate >> 𝐴
𝑄ሶ = 𝑅 𝑄
𝑑𝑡 𝐵

= 𝐵𝐴𝑅ሶ 𝐵𝑄 + 𝐵𝐴𝑅 𝐵𝑄ሶ

𝐴𝑄ሶ = 𝐵𝐴𝑅ሶ 𝐵𝑄 = ( 𝐵𝐴Ω × 𝐵𝐴𝑅) 𝐵𝑄

𝐴𝑄ሶ = 𝐵𝐴Ω × 𝐴𝑄

𝑄ሶ = Ω × 𝑄
12
Time Varying Position and Orientation
Linear and Rotational Velocity of Rigid Bodies
f. Rotational Velocity
• Simultaneous linear and rotational velocity ( 𝑩𝑽𝑸 ≠ 𝟎)
The orientation of frame {B} wrt frame {A} is changing in time and denoted by 𝐵𝐴Ω

• General Velocity

Linear Velocity Linear Velocity Rotational

of {B} of point Q in {B} velocity of {B}

13
Jacobian
Motions of the links of a robot
a. Velocity ‘propagation’ from link to link
• At any instant, velocity of any link has some linear and angular velocity e.g. Link ‘i’
• As frame is rigidly attached to link, link velocity means the frame velocity

• Velocity of link ‘i+1’ will be that of link ‘i’, plus velocity components added by joint ‘i+1’.
• Angular velocity of link ‘i+1’ same as link ‘i’ plus rotational velocity at joint ‘i+1’

14
Jacobian
Motions of the links of a robot
a. Velocity ‘propagation’ from link to link

Description of link i+1 wrt frame {i+1}

Equations of linear velocity of origin of the frame {i+1} is same as {i} plus rotational velocity of
link ‘i’

Multiplying both sides by 𝑖+1𝑖𝑅

15
Jacobian
Motions of the links of a robot
b. If joint ‘i+1’ is prismatic then,

• Applying these equations successively from link to link, we can compute NωN and NvN the
rotational and linear velocities of the last link.

• Note that the resulting velocities are expressed in terms of frame {N}. This turns out to be
useful, as we will see later.

• If the velocities are desired in terms of the base coordinate system, they can be rotated into
base coordinates by multiplication with 𝑁0𝑅

16
Jacobian
Motions of the links of a robot
Example

A two-link manipulator with rotational joints is shown in Fig. Calculate

the velocity of the tip of the arm as a function of joint rates. Give the
answer in two forms - in terms of frame {3} and also in terms of frame
{0}.

17
Jacobian D-H Parameters
Link 𝜶𝒊−𝟏 𝒂𝒊−𝟏 𝒅𝒊 𝜽𝒊
1 0 0 0 𝜃1
2 0 L1 0 𝜃2
3 0 L2 0 0

18
19
Jacobian

20
Jacobian
Jacobian
It is a multidimensional form of the derivative ex. If we have six functions, each of which is a
function of six independent variables,

21
Jacobian

22
Jacobian

23
Jacobian
• In the field of robotics, we generally use Jacobians that relate joint velocities to Cartesian velocities
of the tip of the arm—for example

where ϴ is the vector of joint angles of the manipulator and 𝑣 is a vector of cartesian velocities
• For the general case of a six-jointed robot, the Jacobian is 6 × 6, ϴሶ is 6 × 1, and 0𝑣 is 6 × 1. This
6 × 1 cartesian velocity vector is the 3 × 1 linear velocity vector and the 3 × 1 rotational velocity
vector stacked together:

• Jacobians of any dimension (including nonsquare) can be defined.

• The number of rows equals the number of degrees of freedom in the Cartesian space
• The number of columns in a Jacobian is equal to the number of joints of the manipulator.

24
Jacobian
• In the case of a two-link arm, we can write a 2 × 2 Jacobian that relates joint rates to end-effector
velocity.
• From the result of Example 5.3, we can easily determine the Jacobian of our two-link arm. The
Jacobian written in frame {3} is

• Jacobian written in frame {0} is

25
Jacobian
Changing Jacobian’s Frame of Reference
Given: • This clearly shows the method of
transforming Jacobian from one frame of
reference to another
As the velocity transformation can be given as:

So we can write:

26
Singularities
• As we have a linear transformation relating joint velocity to Cartesian velocity, a reasonable
question to ask is: Is this matrix invertible?

• We can invert it to calculate joint rates from given Cartesian velocities:

• Is the Jacobian invertible for all values of ϴ? If not, where is it not invertible?

• Most manipulators have values of ϴ where the Jacobian becomes singular.

• Such locations are called singularities of the mechanism or singularities for short.

• All manipulators have singularities at the boundary of their workspace, and most have loci of
singularities inside their workspace

27
Singularities
• Workspace-boundary singularities occur when the manipulator is fully stretched out or folded
back on itself in such a way that the end-effector is at or very near the boundary of the workspace

• Workspace-interior singularities occur away from the workspace boundary; they generally are
caused by a lining up of two or more joint axes

• When a manipulator is in a singular configuration, it has lost one or more degrees of freedom (as
viewed from Cartesian space).

• This means that there is some direction (or subspace) in Cartesian space along which it is
impossible to move the hand of the robot, no matter what joint rates are selected.

• It is obvious that this happens at the workspace boundary of robots.

28
Singularities
Example 5.4
Where are the singularities of the simple two-link arm of Example 5.3? What is the physical
explanation of the singularities? Are they workspace-boundary singularities or workspace-interior
singularities?

Where the determinant is equal to zero, the Jacobian has lost full rank and is singular:

Clearly, a singularity of the mechanism exists when θ2 is 0 or 180 degrees

- when θ2 = 0, the arm is stretched straight out
- when θ2 = 180, the arm is folded completely back on itself

These singularities are workspace-boundary singularities, because they exist at the edge of
the manipulator’s workspace
29
Singularities
The danger in applying ϴሶ = 𝐽−1 (ϴ)𝑣 in a robot control system is that, at a singular point, the inverse
Jacobian blows up!
This results in joint rates approaching infinity as the singularity is approached

Example 5.5
Consider the two-link robot from Example 5.3 as it is moving its end-effector along the 𝑋෠ axis at 1.0
m/s, as in Fig. 5.10. Show that joint rates are reasonable when far from a singularity, but that, as a
singularity is approached at θ2 = 0, joint rates tend to infinity.

30
Singularities
We start by calculating the inverse of the Jacobian written
in {0}:

Then, using above eq for a velocity of 1 m/s in the 𝑋෠ direction, we can calculate joint rates as a
function of manipulator configuration:

Clearly, as the arm stretches out toward θ2 = 0, both joint rates go to infinity

31
Static Forces in Manipulators
• Typically, the robot is pushing on something in the environment with the chain’s free end (the end-
effector) or is perhaps supporting a load at the hand.

• We wish to solve for the joint torques that must be acting to keep the system in static equilibrium.

• In considering static forces in a manipulator, we first lock all the joints so that the manipulator
becomes a structure. We then consider each link in this structure and write a force-moment balance
relationship in terms of the link frames.

• We compute what static torque must be acting about the joint axis in order for the manipulator to
be in static equilibrium

• In this way, we solve for the set of joint torques needed to support a static load acting at the end-
effector

32
Static Forces in Manipulators
• The static forces and torques we are considering at the joints are those caused by a static force or
torque (or both) acting on the last link—for example, as when the manipulator has its end-effector
in contact with the environment
• We define special symbols for the force and torque exerted by a neighbor link:
fi = force exerted on link i by link i − 1,
ni = torque exerted on link i by link i − 1.

33
Static Forces in Manipulators

34
Static Forces in Manipulators

35
Static Forces in Manipulators
• What torques are needed at the joints in order to balance the reaction forces and moments acting on
the links?
• All components of the force and moment vectors are resisted by the structure of the mechanism
itself, except for the torque about the joint axis.
• Therefore, to find the joint torque required to maintain the static equilibrium, the dot product of the
joint-axis vector with the moment vector acting on the link is computed:

• In the case that joint i is prismatic, we compute the joint actuator force as

As a matter of convention, we generally define the positive direction of joint torque as the direction
which would tend to move the joint in the direction of increasing joint angle

36
Static Forces in Manipulators
EXAMPLE 5.7
The two-link manipulator of Example 5.3 is applying a force vector 3F with its end-effector.
(Consider this force to be acting at the origin of {3}.) Find the required joint torques as a function of
configuration and of the applied force. (See Fig.)

We apply Eqs. (5.80) through (5.82), starting from

the last link and going toward the base of the robot:

37
Static Forces in Manipulators

It is not a coincidence that this matrix is the transpose of the Jacobian that we found in (5.66) 38
Jacobian in Force Domain
We can equate the work done in Cartesian terms with the work done in joint-space terms. In the
multidimensional case, work is the dot product of a vector force or torque and a vector displacement.
Thus, by principle of virtual work, we have

The definition of the Jacobian is

Therefore

Transposing both sides yields this result:

39
Cartesian Transformations of Velocities and Static Forces

z
z

Now, (5.100) relates velocities in one frame to those in another, so the

6 x 6 operator will be called a velocity transformation;

40
Cartesian Transformations of Velocities and Static Forces

41