MODELLING OF MOBILE ROBOT DYNAMICS

Edouard Ivanjko1 , Toni Petrinic2 , Ivan Petrovic1

1

University of Zagreb, Faculty of Electrical Engineering and Computing

10000 Zagreb, Unska 3, Croatia
2
HEP d.d., 10000 Zagreb, Ulica grada Vukovara 37, Croatia
[email protected](Edouard Ivanjko)
Abstract

This paper presents two approaches to modelling of mobile robot dynamics. First
approach is based on physical modelling and second approach is based on experimental identification of mobile robot dynamics features. Model of mobile robot
dynamics can then be used to improve the navigational system, especially path
planing and localization modules. Localization module estimates mobile robot
pose using its kinematic odometry model for pose prediction and additional sensor measurements for pose correction. Kinematic odometry models are simple,
valid if mobile robot is travelling with low velocity, low acceleration and light
load. Disadvantage is that they dont take any dynamic constraints into account.
This leads to errors in pose prediction, especially when significant control signal
(translational and rotational velocity reference) changes occur. Problem lies in the
fact that mobile robot cant immediately change its current velocity to the desired
value and mostly there exists a communication delay between the navigation computer and mobile robot micro-controller. Errors in predicted pose cause additional
computations in path planning and localization modules. In order to reduce such
pose prediction errors and considering that mobile robots are designed to travel
at higher velocities and perform heavy duty work, mobile robot drive dynamics
can be modelled and included as part of the navigational system. Proposed two
modelling approaches are described and first results using a Pioneer 3DX mobile
robot are presented. They are also compared regarding to complexity, accuracy
and suitability of implementation as part of the mobile robot navigational system.
Keywords: Modelling, mobile robot, estimation, dynamic model.
Presenting Authors Biography
Edouard Ivanjko received his PhD in 2009. from Faculty of Electrical
Engineering and Computing, University of Zagreb, Croatia where he is
currently working. His main research interests are: mobile robotics, localization of mobile robots and environment modelling for mobile robots.
He published 3 papers in international journals and 19 papers in proceedings of international conferences. He is a member of IEEE, KoREMA and
Croatian Society for Robotics. He speaks fluent German and English.

1 Introduction
In recent years huge interest in autonomous industrial
vehicles can be noticed. Control systems for such vehicles should take into account all disturbances which
can appear during their missions. Their control systems
should react quickly and adapt to changing environment
conditions. Many classical methods for designing
control systems (optimum-control methods, algebraic
methods) for such vehicles require a physical system
description together with all of its parameters. For accurate mobile robot system description its necessary to
generate an appropriate dynamic model. The dynamic
model allows considering such properties as: mass, inertia, friction forces, centrifugal force, torque, etc. Such
models are built in order to better understand the structure and operation of the controlled mechatronic system. Creation of such a model becomes even more important if high complex systems have to be developed.
Preparation and application of needed dynamical model
allows early detection of flaws and mistakes in the description (model) of the real system. Their modification
is simpler and less expensive in the virtual construction
stage in comparison with physical prototypes.
Knowledge about mobile robot dynamics is very important for planning feasible mobile robot trajectories.
First a path consisting of a series of poses is generated
in the path planning module. Such a path is generated
regarding to mobile robot dimensions in the sense that
the mobile robot can traverse free workspace underlying planned path. In this case dynamic properties are
not important. Desired path following is only done in
geometric space and path planning criterion is that the
mobile robot doesnt collide with any stationary obstacles. Typical path planning algorithms for such applications are the A* and D* algorithms [1].
Things change when moving obstacles are part of mobile robot workspace. They can be people or other mobile robots or other moving machines. In these situations a certain pose must be reached by the mobile robot
in a certain time frame to avoid collision. In such cases
mobile robot motion velocity has to be constantly altered in an appropriate way that mobile robot can avoid
collision with a moving obstacle. Only in that way a
collision free motion in a workspace with moving obstacles can be guaranteed [2]. During trajectory planning phase mobile robot dynamic properties are used to
create a set of velocity profiles that can be performed by
the mobile robot in a safe manner. In this process mobile robot velocity limitations must be respected with
preserved trajectory curvature [3].
Mobile robot dynamical model is very important in
cases when mobile robot velocities generate forces
which influence cant be neglected during movement.
A good example is mobile robot soccer where used mobile robots are small and velocities are significant compared to their size and mass. Typical shape is a cube
of about 7.5 (cm) size and velocities can be faster than
2 (m/s). Small size and mass combined with such velocity values can cause slippage of drive wheels or turn
overs in a curvature trajectory. Slippage can especially

occur when significant velocity changes (big acceleration or deceleration values) are requested from the mobile robot. Appropriate control strategies and trajectories need to include mobile robot dynamics properties [4].
Recent published research also uses mobile robot dynamics to cope with workspace floor characteristics.
In [5] authors developed a two-level mobile robot
motion control strategy that can cope with different
workspace floor characteristics. Low level part is a classical wheel velocity controller whereas high level part
uses measured wheel velocities to adapt generated trajectory if there are significant differences from desired
values. For this purpose authors modelled the mobile
robot as a rigid body that rolls on two drive wheels and
one castor including velocity changes constraints by the
used actuators.
Mobile robot dynamic model can be also used to improve estimated pose in the localization module. This
module mostly uses a kinematic model to predict mobile robot pose using appropriate control input values.
In this case control input consists of translational and
rotational velocities. Predicted pose is then corrected
using additional sensor measurements. A typical example is usage of non-linear Kalman filters for this
task [6]. Quality of so estimated pose depends on used
kinematic model accuracy, used additional sensors and
workspace model (map) accuracy. Kinematic model accuracy has a significant influence in such a framework.
Control input for kinematic model is usually computed
in a navigational computer connected via a communication link with a low level micro-controller. The
micro-controller handles drive wheels velocity control
and their current velocity measurement. Velocity measurement is then returned to the navigational computer
and used for pose estimation or path planning computations. Control input values of the navigational computer can differ from values that the micro-controller
currently uses for drive wheels velocity control. Reasons are communication link delay, mobile robot dynamic properties and used drive mechanical characteristics. Mechanical characteristics include influence of
friction, backlash, etc. Result is that used kinematic
model predicts mobile robot movement, while microcontroller hasnt received movement command yet or
predicts a mobile robot movement that cant be performed by the used mobile robot. So, there exists a constant prediction error that cant be taken into account by
means of calibration or path planning restrictions.
It would be beneficial to model such features and include them in the pose prediction step as part of the
localization module. Also path planning could be improved, especially the moving obstacle avoidance part
when generated path is altered in time space i.e. a
trajectory is created. To do this mobile robot physical properties have to be examined and their influence
on interesting variables has to be determined. In case
of mobile robot navigation, crucial variable is its pose
which is predicted i.e. estimated using drive wheel velocities. Input variables are velocity references and ap-

propriate model should use them too as an input to provide current mobile robot velocity value on its output.
These velocity values can then be used by the localization and other modules. This would be the first step or
model creation. Second step would include validation
of obtained model. It isnt good to validate the model
on real mobile robot in its working environment due to
danger of damage. More preferable is simulation testing where velocity data from mobile robot experiments
are used. Such a way is used in this article also.
This paper presents two approaches to modelling dynamic mobile robot features including influence of mechanical drive characteristics. First approach is based
on making a physical model of mobile robot body and
components used for velocity control like velocity controller, motor, gearbox, etc. Second approach is based
on experimental fitting of recorded mobile robot velocity data regarding reference velocity data. Both models
are validated using velocity data recorded using a Pioneer 3DX mobile robot.
The rest of the paper is organised as follows. Second section contains description of both modelling approaches. Third section gives an overview on models
implementations in M ATLAB/S IMULINK. After that
obtained results are given followed with models comparison. Paper ends with conclusion and description of
future work on this topic.

2 Modelling approaches
As mentioned above, this article presents two approaches to mobile robot dynamics modelling. First
steps needed for physical modelling are described and
secondly steps taken for experimental identifications
are described. Both models need to take into account
maximal values of rotational and translational velocities including maximal rotational and translational acceleration and deceleration values. These model features can be easily taken into account by using saturation and slope limitation functions.
2.1 Physical modelling
Used Pioneer 3DX robot is a two wheeled differential drive robot, where each wheel is driven independently. Forward motion is produced by both wheels
driven at the same rate, turning right is achieved by
driving the left wheel at a higher rate than the right
wheel and vice versa for turning left. This type of mobile robot can turn on the spot by driving one wheel
forward and second wheel in the opposite direction
at same rate. Third wheel is a castor wheel needed
for mobile robot stability. Drive wheels are equipped
with encoders and their angular velocity readings become available through simple routine calls. Kinematic
model of a differential drive mobile robot can be found
in [7] and geometrical dependencies are given in Fig. 1,
where r is drive wheel radius (mm), vL and vR are left
and right drive wheel velocities, respectively (mm/s),
x and y present mobile robot position in cartesian coordinates in (mm), and b is axle length between drive
wheels (mm).

left actuated
wheel

center of mass

center of axle

castor wheel

right actuated
wheel

Fig. 1 Geometrical dependencies of a differential drive

mobile robot.

Dynamic motion equation can be derived using EulerLagrange formulation [8, 9]:
d
dt

L
qi

L
= Qi ,
qi

(1)

where L stands for difference of kinetic, T , and potential, U , energy, qi stands for generalized coordinate, and
Qi stands for generalized force that acts on the mechanical system.
Under assumption that mobile robot moves only on a
plane surface, potential energy of robot is zero (U = 0)
and we have to find only kinetic energy of the mobile
robot. Kinetic energy of the whole structure is given by
the following equation:
T = Tt + Tr + Trwr ,

(2)

where Tt is kinetic energy of mobile robot translation,

Tr is kinetic energy of mobile robot rotation, and Trwr
is kinetic energy of rotation of wheels and rotors of DC
motors, all in (kgmm2 /s2 ) respectively. Values of introduced energy terms can be expressed by following
equations:
Tt =

1
1
mv 2 = m(x2c + yc2 ),
2 c
2

(3)

1 2
IA ,
2

(4)

Tr =

Trwr =

1 2
1 2
I0 R + I0
L,
2
2

(5)

where m is the mass of entire mobile robot (kg), vc

is linear velocity of the mobile robots center of mass
C (mm/s), IA is the moment of inertia of entire mobile robot with respect to the point A (kgmm2 ), is
is mobile robot romobile robot orientation (rad),
tational speed (rad/s), I0 is the moment of inertia of
combined drive motor (rotor) and wheel (kgmm2 ), and

 R , and
L are angular velocities of the right and left

drive wheel, respectively (rad/s).

R + A
L = ML K
L,
B

Components of the velocity of point A, can be ex R and

L:
pressed in terms of

(
A=

x A = (
R + L ) cos(),
2

(6)

r
L ) sin(),
(R +
2

(7)

y A =

)
(
)
IA + md2 r2
mr2
+
+ I0 ,
4
b2
(

B=
2.2

= r(R L ) ,

(8)

where x A presents velocity of point A in direction of

the X-axis (mm/s), and y A presents velocity of point
A in direction of the Y -axis (mm/s).
Components of the velocity of point C considering velocity of point A are now:
sin ,
x C = x A + d

(9)

cos ,
y C = y A d

(10)

where d is the distance between points A and C in

(mm), and x C presents velocity of point C in direction of the X-axis (mm/s), and y C presents velocity
of point C in direction of the Y -axis (mm/s).
Total kinetic energy of the mobile robot can be calcu R and
L:
lated in terms of

(15)

(
) )
IA + md2 r2
mr2

.
4
b2

(16)

(17)

Experimental identification

Number of needed values is much smaller in this case

and it consist of maximal rotational and translation velocity including their maximal acceleration and deceleration values.
Other features can be obtained by creating appropriate experiments. To obtain this features, first critical
velocity change cases have to be defined. Such cases
are partly covered with the above mentioned maximal
values. Other cases are, when mobile robot changes
its translational or rotational velocity direction, when
it starts its movement (step wise velocity reference
change), and when velocity is constant (steady movement) or constantly changing (mobile robot is accelerating or decelerating). Figs. 2 to 5 show mobile robot
reference and measured velocity relationship for mentioned cases. Only translational velocities are presented
whence rotational velocity responses show similar behaviors. Velocity responses where obtained using a Pioneer 3DX mobile robot controlled with an application
using 100 (ms) sampling time. Only for stand still area
observation smaller sampling time was used (20 (ms)).
150

R ,
L)
T (

(IA +md2 )r 2
mr 2
8 +
2b2

(
+
(
+

)
+

(IA +md2 )r 2
2b2

mr 2
8

mr 2
4

(I +md2 )r 2
A b2

I0
2

reference value
measured value

R
)

I0
2

)
R
L.

(11)

translational velocity [mm/s]

100

50

Now, Lagrange equations:

d
dt
d
dt

L
R

L
L

L
R,
= MR K
R

(12)

L
L,
= ML K
L

(13)

0.5

1.5

time [s]

are applied. MR and ML are right and left actua R and

tion torques, respectively in (kgmm/s2 ) and K

K L are viscous friction values of right and left wheelmotor systems, respectively in (kgmm/s2 ). Finally,
dynamic motion equations can be expressed as:
R + B
L = MR K
R,
A

(14)

Fig. 2 Mobile robot velocity response in case of a steplike velocity reference change.
It can be seen that in case of a step-like velocity reference change (Fig. 2), mobile robot velocity response is
similar to a ramp determined by maximal acceleration
value. Such behavior is expected but there is also an
additional time delay in the velocity response. It can
be explained as a communication delay. Fig. 3 displays
case of a constant reference. Mobile robot can hold desired velocity with influence of noise. On the right side

cluding a greater error when the ramp begins. One part

of the greater error happens due to used drive controller
influence and friction. Fig. 5 displays the case of a velocity direction change. An area where mobile robot
stands still can be observed. Deceleration to stand still
happens constantly like the reference but acceleration in
the opposite direction continues after velocity reference
reaches a certain value. This feature can be explained
with influence of movement in previous direction and
friction.

translational velocity [mm/s]

140
reference value
measured value
145

150

155

160
3

7
time [s]

10

11

Fig. 3 Mobile robot velocity response in case of a constant velocity reference.

35
reference value
measured value

translational velocity [mm/s]

30

25

20

15

10

0
13

13.1

13.2

13.3
time [s]

13.4

13.5

13.6

Fig. 4 Mobile robot velocity response in case of a ramplike velocity reference change.

30
reference value
measured value

translational velocity [mm/s]

20

In order to reduce error between control input values

used for velocity values prediction and true mobile
robot movement values, appropriate dynamic model
can be used. It has to take into account all mentioned
cases. Additionally, model has to be simple so that it
doesnt make any additional burden on the navigation
computer. Also a small number of model parameters
is preferable to enable a possibility of their on-line estimation. Communication delay can be modelled using transport delay, and behavior around standstill using a variable threshold. After reference value reaches
threshold value, estimated velocity values start to rise.
Crucial model values that have to be obtained experimentally are communication delay and threshold at zero
velocity value. Communication delay value is determined by comparing sent velocity reference value and
measured velocity value in time space. Accordingly,
each reference and velocity value had a corresponding
time stamp. Obtained time delay value is 250 (ms) for
translational part and 270 (ms) for rotational part. It
has to be mentioned here that mobile robot controller
works internally with a sampling time of 5 (ms), and
sends out averaged velocity values every 100 (ms).
This explains why obtained communication delay value
isnt an integer multiple of the sampling time. For this
reasons communication delay is modelled as a combination of integer delay and ZOH discretized first order
transfer function. That means that delay of 270 (ms)
in case of the translational part is modelled as delay of
two discretization steps and the rest of 70 (ms) as a
first order discrete transfer function with time constant
of 100 (ms) and discretization time of 100 (ms).
Needed threshold value depends on the ramp slope, i.e.
velocity acceleration or deceleration and its value is
then computed from a dependency recorded from a series of experiments. Experiments where done for characteristic values of velocity change and a good enough
approximation can be made using a straight line. Fig. 6
displays obtained threshold values for translational part
of the model. Rotational part is obtained in an analog
way.

10

10

20

30
25.8

26

26.2

26.4
time [s]

26.6

26.8

27

Fig. 5 Mobile robot velocity response in case of a ramplike velocity reference change crossing stand still point.

of the figure, influence of the mentioned time delay can

be observed. Fig. 4 displays case of a ramp-like velocity
reference change. Also a time delay can be observed in-

Model implementation

This section describes how obtained models were implemented in M ATLAB/S IMULINK. Both models are
implemented with equal requirements. First requirement is of course best possible accuracy. Second requirement was that model can accept measurements obtained from a real mobile robot and compare velocity
values for model accuracy validation. For the sake of

threshold value arround stand still point [mm/s2]

16

ometric center of that element. This is a good approximation because all elements have a symmetrical shape
and constitution like batteries, wheels, motor with gearbox, case, etc. PI controller for drive wheel angular velocity is used. Integral time constant compensates dominant time constant of the velocity control loop. Proportional gain is chosen in order to damping factor of
regulation loop be satisfied ( = 0.9). PI parameters
were so KR = 11996 and TI = 4.58 (s).

14
12
10
8
measured threshold value
estimated threshold value

6
4
2
0

50

100

150

200

250

300
2

mobile robot acceleration/deceleration value [mm/s ]

Fig. 6 Threshold value for translational part.

comparison, deviation of the estimated velocity from

the measured one is computed including average and
maximal error values.
3.1 Physical model
Pioneer 3DX drive system uses high-speed, hightorque, reversible-DC motors, each equipped with a
high-resolution optical quadrature shaft encoder for
precise position and speed sensing and advanced deadreckoning. Motor parameters can be found in [10] and
most important ones are given in Tab. 2. Mobile robot
parameters, motor gear head ratios and encoder ticks
per revolution can be found in [11] and most important
ones are given in Tab. 1. It has to be noticed that values I0 and IA were estimated and the rest is taken from
mobile robot manufacturer data.

Obtained model can be seen in Fig. 7. Model input

variables are mobile robot rotational and translational
velocity references. Their values and known mobile
robot kinematic model with velocity and acceleration
constrains are then used to compute left and right drive
wheel speed references including time delay. These references are then used as input for left and right drive
wheel speed controller. Coupling between left and right
side is also modelled. End part of the model on the right
side computes final mobile robot rotational and translational velocities. Number of encoder ticks per revolution equals 500.
translation
velocity
reference
[mm/s]

velocity and
acceleration left wheel angular
constrains
velocity reference
+
kinematic
right wheel angular
model
+
velocity reference
time delay

esimated
translation
velocity
[mm/s]
PI
controller

DC
motor

PI
controller

DC
motor

rotation
velocity
reference
[o/s]

Mobile
robot
dynamics
+
encoders

Mobile
robot
kinematic
model

estimated
rotation
velocity
[o/s]

Fig. 7 Block scheme of proposed physical model.

Tab. 1 Mobile robot parameters
Parameter
m
r
b
d
I0
IA

Value
28.05
95
320
57.8
9.24 102
175 103

Unit
(kg)
(mm)
(mm)
(mm)
(kgmm2 )
(kgmm2 )

Tab. 2 Mobile robot drive parameters

Parameter
KA
KM
TA
TM
K

Value
0.013
0.029
1.1
8.5
35 107

Unit
(A/V
)
(N m/ W )
(ms)
(ms)
(N ms/rad)

Coordinates of mobile robot center of mass and moment of inertia with respect to the mobile robot center of mass were computed by separating used mobile
robot on distinctive elements whose mass and pose inside mobile robot could be easily measured. It is assumed that mass of each element is concentrated in ge-

Time delay is in this case difficult to model because

amount of time needed for velocity data transmission
and their evaluation in navigational computer or mobile robot micro-controller isnt documented. It can be
heuristically determined and verified in simulation. So,
time delay of 3 time steps (300 (ms)) was used.
3.2

Experimental model

Fig. 8 presents proposed experimental model block

scheme consisting of parts described in subsection 2.2.
It can be used for translational and rotational velocity
part. Model input is generated velocity reference and
output is estimated mobile robot velocity. To obtain estimated values of both, rotational and translational velocities, two models have to work in parallel.
According to model blocks in Fig. 8, proposed model
can be implemented mostly using standard S IMULINK
blocks. Only problematic part is the block labelled
Threshold at zero velocity. It has to influence the estimated velocity value only when mobile robot velocity
is changing its direction. And then, only in the case
when absolute velocity value is beginning to rise. This
part is solved by detecting appropriate velocity change
situation and then applying one of the following cases:

Velocity reference

Tab. 5 Error values in case of rotational velocity estimation in ( /s) for second test values set

Communication delay
Velocity and
acceleration constrains
Threshold at
zero velocity

Error
Maximal
Average

(i) estimated velocity value isnt changed, (ii) estimated

velocity value is changed to zero. In that way separately
deceleration to mobile robot stand still for the case of
negative and positive velocity are detected. Estimated
velocity changes are detected by comparing previous
and current values. A wrap to zero block is used to
make needed changes to estimated velocity value when
estimated value is below the set threshold value.

4 Obtained results
In order to test proposed dynamic models two data sets
obtained from a real Pioneer 3DX mobile robot were
prepared. First data set consists of clearly separated
critical velocity change cases. Second data set is taken
from a navigation algorithm testing experiment. Navigation algorithm generated velocity references considering current obstacle situation in mobile robot environment and respecting mobile robot motion constraints. It
presents a more realistic situation and is therefore better
situated for described models accuracy comparison.

Experimental model
8.94
0.92

Tab. 6 Error values in case of translational velocity estimation in (mm/s) for second test values set
Error
Maximal
Average

Estimated velocity

Fig. 8 Block scheme of proposed experimental model.

Physical model
6.25
1.19

Physical model
53.96
6.77

Experimental model
32
4.66

Tab. 7 Error values in case of no model for second test

values set in ( /s) and (mm/s)
Error
Maximal
Average

Rotation
12.87
2.13

Translation
48.83
14.76

Obtained values of maximal velocity error and its average value are given in Tabs. 3 to 6. For comparison
Tab. 7 contains error values for the case with no model.
Significant improvement can be observed. Apart error values, obtained velocity responses were also examined. They are given in Figs. 9 and 11. Only a part
of recorded velocity values are presented for the sake of
a better representation.
Figs. 10 and 12 present error between measured velocity and estimated velocity. Fig. 10 presents translational
part for first data set and Fig. 12 rotational part for second data set. Other cases can be presented in a similar
fashion.

150

translational speed [mm/s]

100

Tab. 3 Error values in case of rotational velocity estimation in ( /s) for first test values set
Error
Maximal
Average

Physical model
4.31
0.81

Experimental model
4.59
0.48

50

50
reference
measurement
physical model
experimental model

100

150

Tab. 4 Error values in case of translational velocity estimation in (mm/s) for first test values set
Error
Maximal
Average

Physical model
18.8
3.36

Experimental model
12.4
1.62

For the sake of a better model comparison using deviation values between this two data sets, average velocity
values have to be known. First data set had maximal
translational velocity of 300 (mm/s) and maximal rotational velocity of 50 ( /s). Second data set had maximal values of 600 (mm/s) and 63 ( /s), respectively.

26.8

27

27.2

27.4
time [s]

27.6

27.8

28

Fig. 9 Translational velocity response detail in case of

the first data set.

Model comparison

As it was expected, second data set results with a less

accurate estimation. One reason for the less accurate
estimation is more frequent change of the velocity reference. Second reason are situations that cant be detected by an off-line model like drive wheel slippage or

value can be used as a good base for real mobile robot

travelling velocity estimation in combination with other
sensors [12].

translational speed difference [ mm/s]

6
4
2
0
2
4
6
8
10
12

10

15

20

25

30

35

time [s]

Fig. 10 Estimated translational velocity error for first

test values set and physical model.
6

rotational speed [ /s]

10

12

14

reference
measurement
physical model
experimental model

16

18
81

81.5

82

82.5
time [s]

83

83.5

84

84.5

Fig. 11 Rotational velocity response detail in case of the

second data set.

Conclusion & future work

This paper presents two modelling approaches regarding mobile robot drive dynamics. First approach considers modelling every element of mobile robot drive
system and corresponding control framework. Second
approach models characteristic velocity change cases.
It results in a simpler model that gives smaller average
velocity error.

4
3
rotational speed difference [ /s]

Another crucial aspect of the proposed dynamic models comparison is possibility of their implementation as
part of mobile robot navigational system. Also possibility of making an on-line parameter estimation framework would be a good feature. In both cases only communication delay can be estimated on-line. One has
to consider that only available measurement is mobile
robot velocity. Regarding possibility of implementation
experimental model is simpler and doesnt suffer from
the accuracy loss when discretized.

2
1
0
1
2
3
4
5

From data given in tables above, it can be observed that

experimental model has smaller average error values
which means it mostly better models examined mobile
robot drive dynamics. Maximal error value is larger in
the case of the experimental model. Such spikes are related to cases of significant velocity reference changes.
In such cases internal states define partly velocity response and they are not included in the experimental
model. That presents a drawback of this model. Physical model contains internal states of mobile robot drive
system because it models the controller, DC motor,
gearbox with encoders, and communication delay with
appropriate transfer functions. So it performs better in
cases where this internal states are significantly influencing velocity response. Currently this model doesnt
include friction and backlash influence. Adding these
features could improve this model accuracy but it would
also complicate its implementation as part of mobile
robot navigational system. So its more beneficial to use
a simpler model with equal accuracy and, if possible,
combine it with an on-line parameter estimation framework.

20

40

60

80

100

time [s]

Fig. 12 Estimated rotational velocity error for second

test values set and experimental model.

travelling over uneven floor. Mobile robot drive wheels

are in such situations under random environment influence and their drive wheel rotation velocity doesnt
reflect real mobile robot travelling velocity. But, their

Model development consists of two phases. First phase

consists of mobile robot examination to create an appropriate model concept i.e. modelling phase. This
phase includes also creation of appropriate experiments
so that all dynamic properties can be included in the
model concept. Second phase includes firstly testing
of the model concept i.e. simulation phase. Testing is
done in M ATLAB/S IMULINK simulation using experimental data from the first phase. Additionally data from
a real mobile robot are used for simulation testing. Such
an approach ensures an enough accurate model to be
finally implemented as part of a mobile robot navigational system.
As mentioned both models are tested in M ATLAB /S IMULINK environment using velocity data ob-

tained from real mobile robot in real navigational conditions. First test results confirm an improvement in
comparison to usage of velocity reference values for
mobile robot motion prediction. When used by navigational system, mobile robot pose prediction could be
more accurate and a more precise generated trajectory
following can be assured.
So future work will go into direction of including this
dynamic model into mobile robot navigational system
(localization and path planing module) and expanding
it with an on-line estimation framework.

7 Acknowledgement
This research has been supported by the Ministry of
Science and Technology of the Republic of Croatia under grant No. 036-0363078-3018. Authors wish also
to thank Marija Dakulovic for help in providing needed
velocity data from her navigation experiments for dynamic model testing.

8 References
[1] Anthony Stentz. Optimal and efficient path planning for unknown and dynamic environments. International Journal of Robotics and Automation,
10:89100, 1993.
[2] M. Djakulovic and I. Petrovic. Dynamic window
based approach to mobile robot motion control in
the presence of moving obstacles. In Proceedings
of IEEE International Conference on Robotics
and Automation - ICRA 2007, Roma, Italy, 10-14
April 2007, pages 19861991, 2007.
[3] G. Oriolo, A. D. Luca, and M. Vendittelli. Wmr
control via dynamic feedback linearization: Design, implementacion, and experimental validation. IEEE Transactions on Control Systems Technology, 10(6):835852, 2002.

[4] Gregor Klancar, Borut Zupancic, and Rihard

Karba. Modelling and simulation of a group of
mobile robots. Simulation Modelling Practice and
Theory, 15:647658, 2007.
[5] A. Albagul and Wahyudi. Dynamic modelling and
adaptive traction control for mobile robots. International Journal of Advanced Robotic Systems,
1(3):149154, 2004.
[6] Edouard Ivanjko, Ivan Petrovic, and Mario Vasak.
Sonar-based pose tracking of indoor mobile
robots. AUTOMATIKA - Journal for control, measurement, electronics, computing and communications, 45(3-4):145154, 2004.
[7] Edouard Ivanjko, Ivan Komsic, and Ivan Petrovic.
Simple off-line odometry calibration of differential drive mobile robots. In Proceedings of
16th International Workshop on Robotics in AlpeAdria-Danube Region, pages 164169, Ljubljana,
Slovenia, june 7-9, 2007.
[8] S. M. LaValle. Planning Algorithms. Cambrige
University Press, New York, 2006.
[9] Jasmin Velagic, Bakir Lacevic, and Nedim Osmic.
Motion Planning, chapter Nonlinear Motion Control of Mobile Robot Dynamic Model, pages 529
550. InTech Education and Publishing, june 2008.
[10] Mobile robot drive motors parameters.
www.allmotion.se/files/pdf/1707.pdf.
[11] Mobile
robot
construction
parameters.
http://www.activrobots.com/robots/p2dx.html.
[12] P. Goel, S. I. Roumeliotis, and G.S. Sukhatme.
Robust localization using relative and absolute position estimates. In Proceedings of the IEEE/RSJ
International Conference on Intelligent Robots
and Systems (IROS), 1999.