AUTONOMOUS M OBILE ROBOTS

Lecture Set 03
Mobile Robot Kinematics

Huei-Yung Lin
[email protected]

Robot Vision Lab

Department of Electrical Engineering
National Chung Cheng University

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 1/75
 R EPRESENTING ROBOT POSITION

The total dimensionality of the robot chassis on the plane is three:

• Two for position in the plane
• One for orientation along the vertical axis, which is orthogonal to the plane

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 2/75
 R EPRESENTING ROBOT POSITION

As shown in the figure:

• The global reference frame is given by the inertial basis {XI , YI }.
• The robot position is represented by a point P in the global reference frame, and the
basis {XR , YR } is used as the robot’s local reference frame.
• The position of P in the global refernece frame and the difference between the
global and local reference frames are x, y, θ, respectively.
YI

YR
XR

XI

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 3/75
 R EPRESENTING ROBOT POSITION

The pose and velocity of the robot can be described by

   
x ẋ
ξI =  y  and ξ˙I =  ẏ 
   
θ θ̇

respectively. The mapping between the global reference {XI , YI } and the robot’s local
reference frame {XR , YR } can be accomplished using the orthogonal rotation matrix:
 
cos θ sin θ 0
R(θ) =  − sin θ cos θ 0 
 

0 0 1

The operations for pose and motion of the robot are denoted, respectively, by

ξR = R(θ)ξI and ξ˙R = R(θ)ξ̇I

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 4/75
 F ORWARD K INEMATIC M ODELS
Suppose a differential drive robot has two wheels, each with diameter r, and the distance
between the wheels is 2l.
• Let the point P representing the robot is halfway between the wheels.
• Given r, l, θ, and the spinning speed of each wheel, ϕ̇1 and ϕ̇2 , a forward kinematic
model would predict the robot’s overall speed in the global reference frame:
 
ẋ
ξ˙I =  ẏ  = f (l, r, θ, ϕ̇1 , ϕ̇2 )
 
θ̇
YI

YR
XR

XI

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 5/75
 F ORWARD K INEMATIC M ODELS
Let the robot’s local reference frame as defined in the figure.
If one wheel spins while the other wheel contributes nothing and is stationary, then the
robot will move instantaneously with half the speed (w.r.t. P ):

1 1
ẋr1 = r ϕ̇1 and ẋr2 = r ϕ̇2
2 2

YI YI

YR castor wheel
XR v(t)

θ ω(t) θ

P P

XI XI

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 6/75
 F ORWARD K INEMATIC M ODELS
In a differential drive robot, the velocity is given by the summation:

1 1
ẋR = ẋr1 + ẋr2 = r ϕ̇1 + r ϕ̇2
2 2

Note: How to make the robot spin?

Since neither wheel can contribute to sideways motion in the robot’s reference:

ẏR = 0

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 7/75
 F ORWARD K INEMATIC M ODELS
If the right wheel (wheel 1) forward spins alone, then the robot pivots around the left
wheel (wheel 2) counterclockwise. Since the line speed of this wheel is r ϕ̇1 , the
contributed angular velocity at P is given by

r ϕ̇1
ω1 =
2l

Similarly, forward spin of the left wheel results in clockwise rotation at P

r ϕ̇2
ω2 = −
2l

Thus, the kinematic model for the differential-drive robot is

  r ϕ̇1 r ϕ̇2

cos θ sin θ 0 2 + 2
ξ̇I = R(θ)−1 ξ˙R =  − sin θ cos θ 0  0 (1)
  

r ϕ̇1 r ϕ̇2
0 0 1 2l − 2l

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 8/75
 W HEEL K INEMATIC C ONSTRAINTS
The first step to a kinematic model of the robot is to express constraints on the motions of
individual wheels.
The motions of individual wheels can be later combined to compute the motion of the
robot as a whole.
We will consider the four basic wheel types:
• Fixed standard wheel
• Steered standard wheel
• Castor wheel
• Swedish wheel
Some assumptions:
• The plane of the wheel always remains vertical.
• One single point of contact between the wheel and the ground plane.
• There is no sliding at the single point of contact. (The wheel undergoes motion only
under conditions of pure rolling and rotation about the vertical axis through the
contact point.)

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 9/75
 W HEEL K INEMATIC C ONSTRAINTS
Under the above assumptions, there are two constraints for each wheel type:
• Rolling contact – the wheel must roll when motion takes place in the appropriate
direction.
• No lateral slippage – the wheel must not slide orthogonal to the wheel plane.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 10/75
 F IXED S TANDARD W HEEL
The fixed standard wheel has no vertical axis of rotation for steering.
• Its angle to the chassis is fixed.
• Possible motion:
• Move back and forth along the wheel plane.
• Rotate around its contact point with the ground plane.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 11/75
 F IXED S TANDARD W HEEL
The position of A is expressed in polar coordinates by distance l and angle α.
The angle of the wheel plane relative to the chassis is denoted by β.
The radius of the wheel is r, and its rotational position around it horizontal axle is a
function of time t : ϕ(t).
YR
.
yR β
α+β

α+β
ϕ, r .
xR
Robot Chassis A α+β
l

α α+β v
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 12/75
 T HE ROLLING C ONSTRAINT
All motion along the direction of the wheel plane must be accomplished by the appropriate
amount of wheel spin so that there is pure rolling at the contact point.
Thus, from velocity along the motion direction in the local reference frame:

ẋR sin(α + β) − ẏR cos(α + β) − θ̇R l cos β = v = r ϕ̇ (2)

YR
.
yR β
α+β

α+β
ϕ, r .
xR
Robot Chassis A α+β
l

α α+β v
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 13/75
 T HE ROLLING C ONSTRAINT
The above equation can be written as
 
' ( ẋR
sin(α + β) − cos(α + β) −l cos β  ẏR  = r ϕ̇
 
θ̇R

in robot’s local reference frame, and

' (
sin(α + β) − cos(α + β) −l cos β R(θ)ξ̇I − r ϕ̇ = 0 (3)

in global reference frame.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 14/75
 T HE S LIDING C ONSTRAINT
The component of the wheel’s motion orthogonal to the wheel plane must be zero:

ẋR cos(α + β) + ẏR sin(α + β) + θ̇R l sin β = 0

YR
.
yR β
α+β

α+β
ϕ, r .
xR
Robot Chassis A α+β
l

α α+β v
P XR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 15/75
 T HE S LIDING C ONSTRAINT
The above equation can be written as
 
' ( ẋR
cos(α + β) sin(α + β) l sin β  ẏR  = 0
 
θ̇R

in robot’s local reference frame, and

' (
cos(α + β) sin(α + β) l sin β R(θ)ξ̇I = 0 (4)

in global reference frame.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 16/75
 E XAMPLE
For the case shown in the figure, wheel A is in a position with α = 0, β = 0.
If θ = 0, then the sliding constraint Eq. (4) reduces to
  
' ( 1 0 0 ẋ
1 0 0  0 1 0   ẏ  = 0 or ẋ = 0
  
0 0 1 θ̇
YR

.
yR

.
l xR
XR
P A

v
Robot Chassis
Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 17/75
 S TEERED S TANDARD W HEEL
The only difference between the steered standard wheel and fixed standard wheel:
• The standard steered wheel may rotate around a vertical axis passing through the
center of the wheel and the ground contact point.
• The steered standard wheel has one more degree of freedom.
• The angle β is a function of time: β(t), for the steered standard wheel.
Thus, the rolling and sliding constraints are
' (
sin(α + β) − cos(α + β) −l cos β R(θ)ξ˙I − r ϕ̇ = 0
' (
cos(α + β) sin(α + β) l sin β R(θ)ξ˙I = 0

β̇ does not have a direct impact on the instantaneous motion constraints of the robot:
• It is only by integrating over time that changes in steering angle can affect the
mobility of a vehicle.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 18/75
 C ASTOR W HEEL
The vertical rotation axis in a castor wheel does not pass through the ground contact point.
The wheel contact point is at position B, connected by a rigid rod AB of length d to A.
The point A has a position specified in the robot’s reference frame.
The castor wheel has two parameters that vary as a function of time:
• ϕ(t) represents the wheel spin over time,
• β(t) denotes the steering angle and orientation of AB over time.
.
yR
v
YR

.
xR
d B
β(t)
ϕ, r

Robot Chassis A
d l

α
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 19/75
 T HE ROLLING C ONSTRAINT
The rolling constraint of the castor wheel is identical to that of the steered standard wheel:
' (
sin(α + β) − cos(α + β) −l cos β R(θ)ξ̇I − r ϕ̇ = 0 (5)

since the offset axis plays no role during motion that is aligned with the wheel plane.
.
yR
v
YR

.
xR
d B
β(t)
ϕ, r

Robot Chassis A
d l

α
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 20/75
 T HE S LIDING C ONSTRAINT
The lateral force on the wheel occurs at point A because this is the attachment point of the
wheel to the chassis.
• Due to the offset ground contact point relative to A, the constraint that there be zero
lateral movement could be wrong.
• Instead, the constraint is much like a rolling constraint – an appropriate rotation of
the vertical axis must take place. .
yR
v
YR

.
xR
d B
β(t)
ϕ, r

Robot Chassis A
d l

α
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 21/75
 T HE S LIDING C ONSTRAINT
Compared to steered standard wheel, two more terms should be added to the sliding
constraint:
• The line velocity due to the angle β of the castor wheel rotation: dβ̇.
• The line veolcity due to the angle θ of the chassis rotation: dθ̇.
.
yR
v
YR

.
xR
d B
β(t)
ϕ, r

Robot Chassis A
d l

α
P XR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 22/75
 T HE S LIDING C ONSTRAINT
Thus, the sliding constraint of the castor wheel is

ẋR cos(α + β) + ẏR sin(α + β) + θ̇R (d + l sin β) + dβ̇ = 0

in the robot’s local reference frame, and

' (
cos(α + β) sin(α + β) d + l sin β R(θ)ξ˙I + dβ̇ = 0 (6)

in the global reference frame.

The above equations mean that
• Any motion orthogonal to the wheel plane must be balanced by an equivalent and
opposite amount of castor steering motion.
• By setting the value of β̇, any arbitrary lateral motion can be acceptable.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 23/75
 T HE S TEERING ACTION
Different results from a steering action of different wheel types (just think the wheel is not
powered!):
• Steered standard wheel: the steering action does not by itseld cause a movement of
the robot chassis.
• Castor wheel: the steering action itself moves the robot chassis because of the offset
between the ground contact point and the vertical axis of rotation.
Omnidirectional motion of the castor wheel:
• Given any robot chassis motion ξ̇I , there exists some value for spin speed ϕ̇ and
steering speed β̇ such that the constraints are met.
• Thus, a robot with only castor wheels can move with any velocity in the space of
possible robot motions.
• Although the kinematics of castor wheels is somewhat complex, such wheels do not
impose any real constraints on the kinematics of a robot chassis.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 24/75
 E XAMPLE OF C ASTOR W HEEL
A real-world example of such a system is the five-castor wheel office chair
• Its castor wheels will spin and steer as needed to achieve any motion without
contact point sliding.
• If each castor wheel houses two motors, one for spinning and one for steering, then
a control system would be able to move the chair along any trajectory in the plane.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 25/75
 S WEDISH W HEEL
Swedish wheel consists of a fixed standard wheel with rollers attached to the wheel
perimeter.
The axes of the rollers are usually antiparallelto the main axis of the fixed wheel
component.
• The Swedish wheel have no vertical axis of rotation, yet are able to perform
omnidirectional motion like the castor wheel.
• One more degree of freedom is provided by the rollers.
• The exact angle γ between the roller axes and the wheel plane can vary.
• Since each axis can spin clockwise or counterclockwise, one can combine any
vector along one axis with any vector along the other axis.
• These two axes are not necessarily independent (except in the case of the Swedish
90-degree wheel).

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 26/75
 S WEDISH W HEEL
The pose of a Swedish wheel is expressed as in a fixed standard wheel, with the addition
of a term, γ, representing the angle between the main wheel plane and the axis of rotation
of the small circumferential rollers.

YR
.
yR
β

α+β+γ−ω = π/2

ϕ, r τ α τ = β+γ
γ γ .
xR
Robot Chassis ω
A
l
α α+β v
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 27/75
 T HE S LIDING C ONSTRAINTS
The instantaneous constraint is due to the specific orientation of the small rollers.
• The axis around which those rollers spin is a zero component of velocity at the
contact point.
• Moving in that direction without spinning the main axis is not possible without
sliding.
Compared to fixed standard wheels, the sliding constraint should be modified by adding
the angle γ such that the motion along the roller axis is zero:

ẋR cos ω + ẏR sin ω − θ̇R l cos τ − r ϕ̇ cos γ = 0

or
ẋR sin(α + β + γ) − ẏR cos(α + β + γ) − θ̇R l cos(β + γ) − r ϕ̇ cos γ = 0
Thus, the sliding constraint in the global reference frame is
' (
sin(α + β + γ) − cos(α + β + γ) −l cos(β + γ) R(θ)ξ˙I − r ϕ̇ cos γ = 0 (7)

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 28/75
 T HE ROLLING C ONSTRAINTS
Orthogonal to the roller axis the motion is not constrained because of the free rotation ϕ̇sw
of the small rollers.
Thus, the rollering constraint is given by
' (
cos(α + β + γ) − sin(α + β + γ) −l sin(β + γ) R(θ)ξ̇I −r ϕ̇ sin γ−rsw ϕ̇sw = 0
(8)
YR
.
yR
β

α+β+γ−ω = π/2
τ τ = β+γ
ϕ, r γ γ α .
xR
Robot Chassis ω
A
l
α α+β v
P XR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 29/75
 90-D EGREE S WEDISH W HEEL
The behavior of the sliding and rolling constraints and thereby the Swedish wheel changes
dramatically as the angle γ varies.
The angle γ = 0 represents the Swedish 90-degree wheel:
• The zero component of velocity is in line with the wheel plane.
• The sliding constraint Eq. (7) reduces to the fixed standard wheel rolling constraint
Eq. (3).
• There is no sliding constraint orthogonal to the wheel plane because of the rollers,
Eq. (8) reduces to
' (
cos(α + β) − sin(α + β) −l sin β R(θ)ξ˙I − rsw ϕ̇sw = 0

• Any desired motion vector can be made to satisfy Eq. (7) by varying the value of ϕ̇.
• Thus, the wheel is omnidirectional.
• It results in decoupled motion: the rollers and the main wheel provide orthogonal
directions of motion.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 30/75
 0-D EGREE S WEDISH W HEEL
The angle γ = π/2 represents the Swedish 0-degree wheel:
• The rollers have axes of rotation that are parallel to the main wheel axis of rotation.
• The sliding constraint Eq. (7) reduces to the fixed standard wheel sliding constraint
Eq. (4).
• The rollers provide no benefit in terms of lateral freedom of motion since they are
simply aligned with the main wheel.
• In this case the main wheel never needs to spin and therefore the rolling constraint
disappears.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 31/75
 S PHERICAL W HEEL
The spherical wheel places no direct constraints on motion.
Such a mechanism has no principle axis of rotation, thus no appropriate rolling or sliding
constraints exist.
The spherical wheel is omnidirectional and places no constraints on the robot chassis
kinematics.
YR
.
yR β
α+β

α+β
ϕ, r .
xR
Robot Chassis A α+β
l

α α+β vA
XR
P

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 32/75
 K INEMATIC C ONSTRAINTS FOR S PHERICAL W HEEL
The roll rate of the ball in the direction of motion vA of point A is given by
' (
sin(α + β) − cos(α + β) −l cos β R(θ)ξ̇I − r ϕ̇ = 0 (9)

Since the wheel rotation orthogonal to the direction of motion is zero, we have
' (
cos(α + β) − sin(α + β) l sin β R(θ)ξ̇I = 0 (10)

The above equations for spherical wheel are exactly the same as for the fixed standard
wheel.
However, the angle β is a free variable for the spherical wheel, but a constant for the fixed
standard wheel.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 33/75
 ROBOT K INEMATIC C ONSTRAINTS
Given a mobile robot with M wheels, how to compute the kinematic constraints of the
robot chassis?
• Each wheel imposes zero or more constraints on robot motion.
• The process is combining all of the kinematic constraints arising from all of the
wheels based on the placement of those wheels on the robot chassis.
The castor wheel, Swedish wheel, and spherical wheel impose no kinematic constraints on
the robot chassis.
• The value ξ˙I can range freely in all of these cases owing to the internal wheel
degrees of freedom.
Only fixed standard wheels and steerable standard wheels have impact on robot chassis
kinematics.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 34/75
 ROBOT K INEMATIC C ONSTRAINTS
When computing the robot’s kinematic constraints, only the fixed standard wheel and
steered standard wheel have to be considered.
Suppose a robot has Nf fixed standard wheels and Ns steerable standard wheels.
• There are totally N = Nf + Ns standard wheels. (There might be some other
non-standard wheels, e.g. omnidirectional wheels.)
• Let βs (t) denote the variable steering angles of the Ns steerable standard wheels.
• Let βf denote the orientation of the Nf fixed standard wheels.
• Let ϕf (t) and ϕs (t) denote the rotational positions of the fixed and steerable
standard wheels.
• Let the aggregate vector
) *
ϕf (t)
ϕ(t) =
ϕs (t)
N ×1

denote the rotational positions of all standard wheels.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 35/75
 T HE ROLLING C ONSTRAINT
From Eqs. (3) and (5), the rolling constraints of all wheels can be collected in a single
expression:
 
ẋR
J1 (βs )  ẏR  − J2 ϕ̇ = 0 or J1 (βs )R(θ)ξ̇I − J2 ϕ̇ = 0 (11)
 
θ̇R

where J2 is a constant diagonal N × N matrix whose entries are radii r of all standard
wheels (including fixed and steerable), J1 (βs ) denotes a matrix with projections for all
wheels to their motions along their individual wheel planes:
) *
J1f
J1 (βs ) =
J1s (βs )
N ×3=(N1f +N1s )×3

Note that J1 (βs ) is only a function of βs , not βf .

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 36/75
 T HE ROLLING C ONSTRAINT
The properties of J1f :
• It is a constant matrix of projections for all fixed standard wheels.
• Its dimension is Nf × 3.
• Each row consists of the three terms in the 1 × 3 row vector from Eq. (3) for each
fixed standard wheel.
The properties of J1s (βs ):
• It is a matrix of size Ns × 3.
• Each row consists of the three terms in the 1 × 3 row vector form Eq. (5) for each
steered standard wheel.
The rolling constraint
J1 (βs )R(θ)ξ̇I − J2 ϕ̇ = 0
describes that all standard wheels must spin around their horizontal aixs an appropriate
amount based on their motions along the wheel plane so that rolling occurs at the ground
contact point.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 37/75
 T HE S LIDING C ONSTRAINT
The sliding constraints of the fixed standard wheel, Eq. (4), and the steerable standard
wheel, Eq. (6), can be combined into a single expression for the sliding constraint for all
standard wheels:
 
ẋR
C1 (βs )  ẏR  = 0 or C1 (βs )R(θ)ξ˙I = 0 (12)
 

θ̇R

where ) *
C1f
C1 (βs ) = (13)
C1s (βs )
N ×3=(Nf +Ns )×3

C1f and C1s are Nf × 3 and Ns × 3 matrices whose rows are the three terms in the 1 × 3
row vector of Eqs. (4) and (6) for all fixed and steerable standard wheels.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 38/75
 T HE S LIDING C ONSTRAINT
The sliding constraint
C1 (βs )R(θ)ξ˙I = 0
describes that, for all standard wheels, the components of motion orthogonal to their wheel
planes must be zero .
The sliding constraint over all standard wheels has the most significant impact on defining
the overall maneuverability of the robot chassis.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 39/75
 K INEMATIC C ONSTRAINTS
Eqs. (11)
J1 (βs )R(θ)ξ̇I − J2 ϕ̇ = 0
and (12)
C1 (βs )R(θ)ξ˙I = 0
can be combined into a single expression:
) * ) *
J1 (βs ) J2 ϕ̇
R(θ)ξ̇I = (14)
C1 (βs ) 0

The above equations relate the robot motion to the rolling and sliding constraints J1 (βs )
and C1 (βs ), and the spin speed of the robot’s wheels, ϕ̇.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 40/75
 D IFFERENTIAL -D RIVE ROBOT
For a differential-drive robot, the unpowered spherical wheel is omnidirectional and does
not enforce any constraints.
Since the remaining two drive wheels are not steerable, J1 (βs ) and C1 (βs ) simplify to J1
and C1 , respectively.

• For the right wheel: α = −π/2, β = π.

• For the left wheel: α = π/2, β = 0.

YI

YR
XR

XI
Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 41/75
 D IFFERENTIAL -D RIVE ROBOT
From Eq. (3), the rolling constraints for both fixed standard wheels can be written as
) * ) *) *
1 0 l r 0 ϕ̇r
R(θ)ξ̇I − =0
1 0 −l 0 r ϕ̇l

From Eq. (4), the identical sliding constraints for both fixed standard wheels can be
written as ' (
0 1 0 R(θ)ξ˙I = 0

The above equations can be combined into the form of Eq. (14):
 
1 0 l ) *
˙ J2 ϕ̇
 1 0 −l  R(θ) ξ = (15)
 
I
0
0 1 0

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 42/75
 D IFFERENTIAL -D RIVE ROBOT
The kinematic equation for the differential-drive robot is given by inverting Eq. (15)
   −1
ẋ 1 0 l ) *
J2 ϕ̇
ξ˙I =  ẏ  = R(θ)−1  1 0 −l 
   
0
θ̇ 0 1 0
 
1 1
2 2 0 ) *
 J2 ϕ̇
= R(θ)−1  0 0 l 

1 1 0
2l − 2l 0

The above equation is the same as Eq. (1) derived previously.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 43/75
 O MNIDIRECTIONAL ROBOT
Consider a robot with three Swedish 90-degree wheels arranged radially symmetrically as
shown in the figure.

YI

v(t)

ω(t)
θ

XI

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 44/75
 O MNIDIRECTIONAL ROBOT
Suppose the robot’s local reference is defined as shown in the figure, the distance between
each wheel and P is l, and all three wheels have the same radius r.

YR
.
yR

.
xR
ω1 1
r P
XR
l

2 ϕ, r
3

ICR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 45/75
 O MNIDIRECTIONAL ROBOT
Since there is no sliding constraint for omnidirectional robot, Eq. (14):
) * ) *
J1 (βs ) J2 ϕ̇
R(θ)ξ̇I =
C1 (βs ) 0

is simplified as
J1 (βs )R(θ)ξ̇I = J2 ϕ̇
It can be further reduced to
ξ˙I = R(θ)−1 J1f
−1
J2 ϕ̇ (16)

since the robot has no steerable wheels.

The entries of J1f is given by the matrix elements of the rolling constraints Eq. (7) for the
Swedish wheel:
' (
J1f = sin(α + β + γ) − cos(α + β + γ) −l cos(β + γ)

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 46/75
 O MNIDIRECTIONAL ROBOT
For the wheel configuration shown in the figure:
• The angle γ = 0 for Swedish 90-degree wheels.
• The angle β = 0 since the wheels are tangent to the robot’s circular body.
π
• The angles α1 = 3 , α2 = π, α3 = − π3 for each wheel.
YR
.
yR YR
.
yR
β
.
xR
α+β+γ−ω = π/2
ω1 1 τ α τ = β+γ
r P γ γ .
XR xR
l Robot Chassis ω
A
2 ϕ, r l
3 α α+β v
P XR

ICR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 47/75
 O MNIDIRECTIONAL ROBOT
From the above setting, we have
   √ 
3
sin π
− cos
3
π
3 −l 2 − 12 −l
J1f = 0 − cos π −l  =  0√ 1 −l 
   
sin(− π3 ) − cos(− π3 ) −l − 23 − 12 −l

Thus,
 
√1 0 − √13
3
ξ˙I = R(θ)−1  − 13 2
− 13  J2 ϕ̇
 
3
1 1 1
− 3l − 3l − 3l

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 48/75
 E XAMPLE
Consider a special case that l = 1, r = 1 for all wheels, the wheel spin speeds
ϕ̇1 = 4, ϕ̇2 = 1, ϕ̇3 = 2, and θ = 0, then
   
ẋ √2
3
ξ˙I =  ẏ  =  − 43
   

θ̇ − 73

That is, the robot can move and turn instantaneously with certain values.
The robot motion can be predicted by combining the rolling constraints of individual
wheels.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 49/75
 M OBILE ROBOT M ANEUVERABILITY
Kinematic mobility of a robot chassis:
• The ability to directly move in the environment.
• The basic constraint is that every wheel must satisfy its sliding constraint.
In addition to instantaneous kinematic motion, a mobile robot is able to further manipulate
its position over time by steering steerable wheels.
Overall maneuverability of a robot:
• The mobility available based on the kinematic sliding constraints of the standard
wheels.
• The additional freedom contributed by steering and spinning the steerable standard
wheels.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 50/75
 D EGREE OF MOBILITY

The sliding constraints Eq. (12):

C1 (βs )R(θ)ξ˙I = 0

must hold for each wheel to avoid any lateral slip.

For the above equation to be satisfied, the motion vector R(θ)ξ̇I must belong to the null
space of the projection matrix
) *
C1f
C1 (βs ) =
C1s

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 51/75
 I NSTANTANEOUS C ENTER OF ROTATION (ICR)
At any given instant, wheel motion along the zero motion line must be zero.
In other words, the wheel must be moving instantaneously along some circle of radius R
such that the center of that circle is located on the zero motion line.
This center is called the instantaneous center of rotation (ICR).
The ICR geometric construction demostrates how robot mobility is a function of the
number of constraints on the robot’s motion, not the number of wheels.

zero motion line

zero motion line

ICR ICR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 52/75
 ICR FOR T WO W HEELS
In the case of bicycle shown in the left figure, each wheel contributes a constraint (or a
zero motion line). Thus, there is only a single point for the ICR.
In the case of differential-drive robot shown in the right figure, the two wheels are aligned
along the same horizontal axis. Thus, the ICR is constrained to lie along a line, not at a
specific point. (Any good?)

zero motion line

ICR

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 53/75
 ICR FOR T WO W HEELS
Although the bicycle and differential-drive chassis have the same number of
non-omnidirectional wheels, the former has two independent kinematic constraints while
the latter has only one:
• For a bicycle, the two constraints are independent and together constrains the
overall robot motion.
• For a differential-drive robot, the second wheel imposes no additional kinematic
constraints on robot motion since its zero motion line is identical to that of the first
wheel.
A wheel may not be able to contribute an independent constraint to the robot kinematics,
for example, the second steerable standard wheel of the Ackerman vehicle. (Why?)

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 54/75
 I NDEPENDENT C ONSTRAINTS FOR ROBOT M OTION
Robot chassis kinematics is a function of the set of independent constraints arising from
all standard wheels.
From (12), the number of independent constraints is given by

rank C1 (βs )

The greater the rank, the more constrained is the mobility of the robot.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 55/75
 ROBOT WITH S INGLE F IXED S TANDARD W HEEL
Consider a robot with exactly one fixed standard wheel at the position α, β, l relative to the
robot’s local reference frame.
There are no steerable standard wheels:
• C1s is empty.
• C1 (βs ) contains only C1f .
• The above matrix has a rank of one since there is only one fixed standard wheel.
• The robot has a single independent constraint on mobility
' (
C1 (βs ) = C1f = cos(α + β) sin(α + β) l sin β

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 56/75
 ROBOT WITH T WO F IXED S TANDARD W HEELS
A robot with two fixed standard wheels has two different configurations:
• Differential-drive robot
• Bicycle with the steering locked in the forward position
For the first case, the matrix C1 (βs ) has two constraints but a rank of one:
) * ) *
cos α1 sin α1 0 cos α1 sin α1 0
C1 (βs ) = C1f = =
cos(α1 + π) sin(α1 + π) 0 − cos α1 − sin α1 0

from Eq. (4), where l1 = l2 , β1 = β2 = 0, α1 + π = α2 .

For the second case, the matrix C1 (βs ) retains two independent constraints and has a rank
of two:
) * ) *
cos(π/2) sin(π/2) l1 sin(π/2) 0 1 l1
C1 (βs ) = C1f = =
cos 3(π/2) sin 3(π/2) l1 sin(π/2) 0 −1 l1

from Eq. (4), where l1 = l2 , β1 = β2 = π/2, α1 = 0, α2 = π.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 57/75
 R ANK OF C ONSTRAINT M ATRIX & D EGREE OF M OBILITY
If rank C1f > 1 then the vehicle can, at best, travel along a circle or along a straight line.
• The robot has two or more independent constraints due to fixed standard wheels that
do not share the same horizontal axis of rotation.
The possible range of rank values for any robot is given by 0 ≤ rank C1 (βs ) ≤ 3.
• rank C1 (βs ) = 0 : It is only possible if there are zero independent kinematic
constraints in C1 (βs ). There are neither fixed nor steerable standard wheels
attached to the robot frame, i.e., Nf = Ns = 0.
• rank C1 (βs ) = 3 : The robot is completely constrained in all directions and is
degenerate.
The degree of mobility δm for a robot is defined as

δm = dim N {C1 (βs )} = 3 − rank C1 (βs )

The dimension of the null space (dim N ) of matrix C1 (βs ) is a measure of the number of
DOF of the robot that can be immediately manipulated through changes in wheel velocity.

0 ≤ δm ≤ 3

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 58/75
 E XAMPLES
For a differential-drive chassis, rank C1 (βs ) = 1 and δm = 2:
• A differential-drive robot can control the change rates in both orientation and its
forward/reverse speed by manipulating wheel velocities.
• Its ICR is constrained to lie on the infinite line extending from its wheels’ axles.
For a bicycle chassis, rank C1 (βs ) = 2 and δm = 1:
• There is one less degree of mobility compared to differential-drive robot.
• A bicycle only has control over its forward/reverse speed by direct manipulation of
wheel velocities.
• Only by steering can the bicycle chage its ICR.
Any robot consisting only of omnidirectional wheels such as Swedish or spherical wheels
will have the maximum mobility, δm = 3.
• Such a robot can directly manipulate all three degrees of freedom.
Note: The degree of mobility quantifies the degrees of controllable freedom based on
changes to wheel velocity.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 59/75
 D EGREE OF S TEERABILITY
The impact of steering is indirect since the robot must move for the change in steering
angle of the steerable standard wheel to have impact on robot pose.
The degree of steerability δs is defined as

δs = rank C1s (βs )

An increase in the rank of C1s (βs ) implies more degrees of steering freedom and thus
greater eventual maneuverability.
Since C1 (βs ) includes C1s (βs ), a steerable standard wheel can both decrease mobility
and increase steerability:
• Its particular orientation at any instant imposes a kinematic constraint.
• Its ability to change that orientation can lead to additional trajectories.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 60/75
 D EGREE OF S TEERABILITY
The range of δs is given by 0 ≤ δs ≤ 2.
• For δs = 0 : The robot has no steerable standard wheels, Ns = 0.
• For δs = 1 : The most common case when a robot consists of one or more steerable
standard wheels.
• For an ordinary vehicle with Ns = 2 and Nf = 2, the fixed wheels share a
common axle and thus rank C1f = 1.
• The fixed wheels and one of the steerable wheels constrain the ICR.
• The second steerable wheel cannot impose any independent kinematic
constraint and thus rank C1s (βs ) = 1.
• In this case, δm = 1 and δs = 1.
• For δs = 2 : Only possible in robots with no fixed standard wheels: Nf = 0.
• The pseudobicycle with two separate steerable standard wheels.
• Its ICR can be placed anywhere on the ground plane, which is also the
meaning of δs = 2.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 61/75
 ROBOT M ANEUVERABILITY
The overall DOF that a robot can manipulate is called the degree of maneuverability δM :

δM = δm + δs

It is defined in terms of mobility and steerability. Thus, the maneuverability consists of:
• The degrees of freedom that the robot manipulates directly through wheel velocity.
• The degrees of freedom that it indirectly manipulates by changing the steering
configuration and moving.

Omnidirectional Differential Omni-Steer Tricycle Two-Steer

δM =3 δM =2 δM =3 δM =2 δM =3
δm =3 δm =2 δm =2 δm =1 δm =1
δs =0 δs =0 δs =1 δs =1 δs =2

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 62/75
 ROBOT M ANEUVERABILITY
Two robots with the same δM are not necessarily equivalent.
Differential-drive and tricycle geometries have equal maneuverability δM = 2:
• In differential drive all maneuverability is the result of direct mobility (2, 0).
• In the case of a tricycle the maneuverability results from steering also (1, 1).
• Neither configuration allows the ICR to range anywhere on the plane.
• In both case, the ICR must lie on a predefined line w.r.t. the robot reference frame.
For any robot with δM = 2 the ICR is always constrained to lie on a line and for any robot
with δM = 3 the ICR can be set to any point on the plane.

Omnidirectional Differential Omni-Steer Tricycle Two-Steer

δM =3 δM =2 δM =3 δM =2 δM =3
δm =3 δm =2 δm =2 δm =1 δm =1
δs =0 δs =0 δs =1 δs =1 δs =2

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 63/75
 S YNCHRO D RIVE C ONFIGURATION
In a three-wheeled synchro drive robot Nf = 0 and Ns = 3, thus rank C1s (βs ) can be
used to determine both δm and δs .
• Since a single motor steers all three wheels using a belt drive, the degree of
steerability δs = 1.
• A fixed steering manipulates only one DOF, the degree of mobility δm = 1.
• Thus, the degree of maneuverability δM = 2 for a synchro drive robot.
A synchro drive robot can only manipulate, in total, two degrees of freedom.
• There is no way for the chassis orientation to change.
• Only the (x, y) position of the chassis can be manipulated.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 64/75
 M OBILE ROBOT W ORKSPACE
For a robot, maneuverability is equivalent to its control degrees of freedom.
We care about the ways in which the robot can use its control degrees of freedom to
position itself in the environment.
For the Ackerman vehicle:
• The total number of control degrees of freedom is δM = 2 : one for steering and the
other for actuation of the drive wheel.
• The total degrees of freedom of the vehicle is 3 : the car can position itself on the
plane at any x, y point and with any angle θ.
• Note: The dimension of a robot’s configuration space can exceed δM .
We also care about how the robot is able to move between various configurations:
• What are the types of paths it can follow?
• What are its possible trajectories through this configuration space?

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 65/75
 V ELOCITY S PACE
Given the kinematic constraints of a robot, its velocity space describes the independent
components of robot motion that the robot can control.
The velocity space of a unicycle can be represented by (ϕ̇, θ̇):
• ϕ̇ represents the instantaneous forward speed of the unicycle.
• θ̇ represents the instantaneous change in orientation.
The number of dimensions in the velocity space of a robot is the number of independently
achievable velocities. It is called the differential degrees of freedom (DDOF).
A robot’s DDOF is always equal to its degree of mobility δm .
• The DDOF of a bicycle is 1 (ϕ̇).
• The DDOF of a omnibot is 3 (ẋ, ẏ, θ̇).

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 66/75
 DOF AND DDOF
The relationship between DOF and DDOF:

DDOF ≤ δM ≤ DOF (17)

A bicycle and an omnibot both have DOF = 3: they can achieve any pose (ẋ, ẏ, θ̇).
• The omnibot can do so by directly achieving the goal positions of all three axes
simultaneously since DDOF = 3.
• The bicycle requires more time since DDOF = 1. (Why?)
The workspace DOF governs the robot’s ability to achieve various poses, and the robot’s
DDOF governs its ability to achieve various paths.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 67/75
 H OLONOMIC ROBOT
A holonomic robot is a robot that has zero nonholonomic kinematic constraints.
A nonholonomic robot is a robot with one or more nonholonomic kinematic constraints.
A holonomic kinematic constraint can be expressed as an explicit function of position
variables only.
• For a mobile robot with a single fixed standard wheel, a holonomic kinematic
constraint would be expressed using α, β, l, r, ϕ, x, y, θ only.
˙
• Such a constraint may not use derivatives of these values, such as ϕ̇ or ξ.
A nonholonomic kinematic constraint requires a differential relationship, such as the
derivative of a position variable.
• It cannot be integrated to provide a constraint in terms of the position variables only.
• Thus, nonholonomic systems are often called nonintegrable systems.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 68/75
 N ONHOLONOMIC C ONSTRAINT
The sliding constraint
' (
cos(α + β) − sin(α + β) l sin β R(θ)ξ̇I = 0

for the fixed standard wheel is non-integrable, depending explicitly on robot motion.
• It uses robot motion ξ˙ rather than pose ξ.
• Thus, the sliding constraint is a nonholonomic constraint.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 69/75
 B ICYCLE
The bicycle with one fixed standard wheel and one steerable standard wheel is a
nonholonomic robot due to the fixed wheel sliding constraint. (Check the previous slide.)
However, the locked bicycle is a holonomic robot:
• It can only move along a curve (or a line).
• The sliding constraint of both wheels can be replaced with an equally complete set
of constraints on the robot pose. (How?)
• This eliminates two nonholonomic constraints, corresponding to the sliding
constraints of the two wheels.
• The rolling constraints can also be replaced with one directly relates position on the
curve with wheel rotation angle ϕ.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 70/75
 H OLONOMIC ROBOT
The first type of holonomic robot:
• The constraints do exist but are all holonomic kinematic constraints.
• This is the case for all holonomic robot withs δM < 3.
• Example: the locked bicycle.
The second type of holonomic robot:
• There do not exist any kinematic constraints, i.e., Nf = 0 and Ns = 0.
• No kinematic constraints imply no nonholonomic kinematic constraints, thus the
robot is always holonomic.
• This is the case for all holonomic robots with δM = 3.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 71/75
 H OLONOMIC ROBOT
A robot is holonomic if and only if DDOF = DOF .
Only through nonholonomic constraints (imposed by steerable or fixed standard wheels)
that a robot can achieve a workspace with DOF exceeding its DDOF, i.e.,
DOF > DDOF .
In mobile robotics, we generally require DOF = 3.
• An omnidirectional robot is a holonomic robot with DDOF = 3.
• The locked bicycle case is out of interest.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 72/75
 PATH AND T RAJECTORY
In mobile robotics, we care about both
• the robot’s ability to reach the required final configurations, and
• how the robot gets there.
For an omnidirectional robot (holonomic in a 3-D workspace)
• Unconstrained wheels should be used, ground clearance and suspension issues.
• Not stable for high speed movement, e.g., rotate with non-negligible centripetal
force.
For a car-like robot (with kinematic constraints)
• Nonholonomic constraints can improve stabilities of movements.
• The lateral centripetal forces are passively counteracted through the sliding
constraints.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 73/75
 T WO -S TEER AND O MNIDIRECTIONAL
A bicycle with both wheels steerable is called the two-steer.
• The degree of steerability is 2, the degree of mobility is 1.
• Thus, the degree of maneuverability δM = 3.
• Although this configuration is not holonomic, it has DOF = 3. (Check Eq. (17).)
• The two-steer can select any ICR, thus can follow any path in its workspace.
Any robot with δM = 3 can follow any path in its workspace from its initial pose to its
final pose.
• Both the two-steer and omnidirectional robots can do so.
• The difference between the DOF granted by steering versus by direct control of
wheel veolcity is path versus trajectory.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 74/75
 PATH AND T RAJECTORY
A trajectory is like a path, except that it occupies an additional dimension: time.
• A path is 3-D (x, y, θ).
• A trajectory is 4-D (x, y, θ, t).
The two-steer requires changes to internal degrees of freedom and because these changes
take time, arbitrary trajectories are not attainable.

Autonomous Mobile Robots - Lecture Set 03 Huei-Yung Lin, National Chung Cheng University – p. 75/75