REMOTE LABORATORY EXPERIMENTS ADDRESSING PATH

PLANNING FOR MOBILE ROBOTS

Grzegorz Zyśko, Florian Zeiger, Klaus Schilling, Markus Sauer

Department of Computer ScienceVII - Robotics and Telematics, Bayerische Julius-Maximilians Universität Würzburg,

Am Hubland D-97074, Wuerzburg, Germany

Keywords: Remote experiment, inverse kinematics, nonholonomic, mobile robot, eLearning.

Abstract: This paper describes an educational remote experiment for path planning with mobile robot hardware which

is accessible via the internet. The experiment uses a nonholonomic car-like mobile robot with an Ackerman-

steering and demonstrates the problems of the inverse kinematics of this kind of mobile robot. It emphasizes

the educational aspects, shows how to combine primitive manoeuvres in order to solve the inverse

kinematics problem, and gives a detailed description of these manoeuvres.

1 INTRODUCTION

Introducing the new media and the internet provides

new possibilities for eLearning. In the last years,

remote laboratories and tele-experiments were

created in order to improve the quality of education.

Many of these remote-experiments are related to

control methods like PI-Controller or the control of

other processes. Providing remote experiments with

mobile robots and especially real hardware

experiments with mobile robots are not so popular.

There are few remote-laboratories working with real

mobile robots (cf. 0/2003a/2003b and 0). (0) uses a

Path Following algorithm, but the experiment is

based on a tracked robot. The experiments in (0 and

2003a are related to the inverse kinematics problem

of a differential drive robot and the PI-Control of the

robot's velocity. These kinds of nonholonomic

mobile robots require less space and the control

methods are easier compared to car-like mobile

robots with an Ackerman steering. 0 presents an

experiment to a nonholonomic motion planner

(NHMP), but this experiment does not use a real car-

like mobile robot. Instead a movable platform to

simulate the nonholonomic behavior of the robot is

use. On that account, this experiment does not

provide the same problems like friction, deviations

because of the mechanical construction, and slippage

like using real car-like mobile robot hardware.

This work presents an experiment of a remote

laboratory which is focused on real hardware

experiments with nonholonomic car-like mobile

robots. The students are starting with experiments of

the forward kinematics where a kinematics model of

the car-like mobile robot is introduced. The

students’ task is to find several variables which are

necessary do build the model e. g. systematic errors

for encoders and gyro, deviations of the mechanical

driving mechanism and accuracy of a fuzzy distance

controller (cf. 0/2004b).Then dynamic and control

aspects are presented in a PI-Controller experiment

(cf. 0/2004b).

The present experiment is a first approach to

build a remote exercise related to Path Planning

methods for nonholonomic car like mobile robots

with Ackerman steering. This Experiment is an

“interface” between an existing forward kinematics

exercise and a new one related to Path planning.

Presented exercise give students the opportunity to

check validity of the created model. The obtained

variables are used to estimate the position of the

robot, after simple or complex motions.

2 EXPERIMENT OUTLINE

The presented remote experiment uses the

experiment area and infrastructure (cf. Fig. 1) of the

remote laboratory for car like mobile robots of the

University of Würzburg shown in 0 and 0.

Performing a remote experiment consists of three

steps. At first, the students get a tutorial. This

431

sko G., Zeiger F., Schilling K. and Sauer M. (2005).

REMOTE LABORATORY EXPERIMENTS ADDRESSING PATH PLANNING FOR MOBILE ROBOTS.

In Proceedings of the Second International Conference on Informatics in Control, Automation and Robotics - Robotics and Automation, pages 431-434

DOI: 10.5220/0001178504310434

 SciTePress

tutorial describes several sub-experiments and gives

basic information to the subject. After students

finished this tutorial, they can pass an online exam

with test questions. Finally the control GUI for the

robot can be accessed and the hardware experiment

described in the tutorial can be performed with our

MERLIN robot (cf. Fig. 3). The design of these

online-experiments is well tested and was evaluated

in previous projects.

The experiment described in this paper consists

of 3 steps, a simulation of different methods of

motion, a test of the simulation results on real

hardware, and finally a demonstration and analysis

of three different path planning approaches.

At first, the students are introduced to the basic

motion of a car-like mobile robot. Afterwards the

students are able to combine basic motions to

primitive maneuvers, like three-point-turn and

sidewise shifting (cf. Section IV). The next step

introduces the students to the combination of

primitive maneuvers to complex maneuvers for

achieving a certain configuration. In order to

highlight issues related to robot maneuverability and

robot control, we assume that obstacle avoidance is

solved by path planning comparisons, and string

concatenations.

Figure 1: Area for the experiments with MERLIN

After the students are familiarized with the

primitive maneuvers (cf. Section IV), a simulation is

provided. Here the inaccuracy of the distance

controller and the influence of this deviation on the

finally achieved pose are demonstrated. These

simulation results are now tested on our real

MERLIN hardware and the students experience the

difference between simulations and the real

hardware. Here students can see the influence of a

deviation of the distance controller in open loop

control and they can analyze the effects of an

orientation error while performing complex

maneuvers. They compare these different maneuvers

with respect to environmental influences and

parameters like the available free space, the required

accuracy, and the traveled distance.

Finally, several path planning algorithms (cf. 0)

like the Road Map, the Potential Fields, and the

Distance Transform methods are presented. The

experiment provides a GUI to the students where

they can examine these three methods and analyze

their usability for different maps. These methods

always return a path that has to be transformed into

segments which are navigable by the used mobile

robots. Therefore we derive a path which consists of

primitive maneuvers. Now, the students can learn

how to combine the methods of path planning and

the different primitive maneuvers in order to move

the robot to a certain configuration.

3 INVERSE KINEMATICS OF

CAR-LIKE MOBILE ROBOTS

Inverse kinematics for nonholonomic mobile robots

with Ackerman-steering is a complex problem due

to the constraints described in Latombe, 1993. A

solution (path) can not be represented by a single

motion or mathematical equation. It was proven 0

that, despite this limitation, the robot remain fully

controllable. For these purposes a combination of

several primitive maneuvers were created. The

experiment provides these maneuvers to the students

for theoretical analysis and tests with real hardware.

1) Primitive Maneuvers: For car-like mobile

robots 0 introduced primitive maneuvers for rotation

and shifting. For the rotation, we use the three-point-

rotation as described in 0 (cf. Fig.3). Here, the path

of the robot consists of two circular arcs of a given

length, a given radius, and a tangential connection

between them. This primitive maneuver enables the

robot to perform clockwise or counter-clockwise

rotations.

For performing a sidewise motion of the robot,

we present the students the typical maneuver for

shifting as it is described in 0. Due to the kinematics

constraints of a mobile robot with an Ackerman-

steering, this maneuver is based on performing

motions on a circle arc with a suitable radius and a

suitable length and moving on a tangential line from

one circle arc to another circle arc (cf. Fig.2).

ICINCO 2005 - ROBOTICS AND AUTOMATION

432

dist

Figure 2: Shifting maneuver (left side) and rotation

maneuver (right side).

One of the common problems of the inverse

kinematics of nonholonomic mobile robots is the

required space. Usually, experiments of inverse

kinematics of mobile robots require a large

experiment area. In Fig.2 we see that the typical

maneuver presented in 0 requires enough space in

each direction of the robot. In a small experiment

area, it could be necessary to modify this maneuver

in a slight way because of the limited space on at

least one side of the robot. Fig. 3 presents two

alternative maneuvers for shifting.

dist

Maneuver A Maneuver B

Figure 3: Alternative methods to shift a nonholonomic

mobile robot

If the robot's starting configuration is located too

close to a wall it is impossible to perform a circular

motion in the direction of the wall. In this case, we

have two possibilities to shift MERLIN.

Maneuvers A and B are still applicable if the

robot is placed close to some obstacles. For the

mathematical representation let R be the minimum

turning radius, d be the wheelbase, and dist be the

distance the robot should be moved sidewise.

2) Complex Maneuvers: To familiarize students

to the problems of the inverse kinematics, we

present two more complex maneuvers to reach a

given configuration.

At first we use the same two primitive

maneuvers like introduced for the differential drive

mobile robot: two rotations and a straight-line

movement. At first the robot performs a three-point

rotation (see Fig. 4b) to the direction of the target

position. Then it moves to the position where the

second rotation should take place (Fig. 4c). After the

second rotation (cf. Fig. 4d), the target pose ptarget

is reached. This approach can use up to seven basic

movements: the rotations consist of three

movements on circle arcs each and the straight-line

movement.

start

target

a) b) c) d)

Figure 4: Rotate-move-rotate maneuver for a

nonholonimic car-like mobile robot

The second approach consists of driving on two

segments of circle arcs and one tangential

connection between them (cf. Fig.5). Using this

method we need a maximum of three basic motions

to reach any target configuration.

start

target

Figure 5: Driving on two circle arcs and their tangential

connections as complex maneuver for navigating to a

target pose

Fig.5 presents only some possible traces from the

start configuration to the end configuration.

Connecting these two pairs of circles with tangential

lines results, always in nine traces to the new

position. Then the most suitable trace can be chosen

with respect to the environmental parameters

4 PATH PLANNING

The hardware experiments are concluded by a

simulation of three basic path planning approaches.

The students are introduced to the Road Map, the

Potential Field, and the Distance Transform method.

The students examine the performance and usability

of these methods with different environmental maps

with varying obstacle arrangements.

So they can decide which path planning method

is the most applicable for the different situations.

REMOTE LABORATORY EXPERIMENTS ADDRESSING PATH PLANNING FOR MOBILE ROBOTS

433

After the usability of the path planning methods is

analyzed, the students have to combine their

knowledge gained in previous experiments. The

provided simulation tool considers a point-robot.

Thus, kinematics constraints are not included and

the resulting path of this path planning method will

not be navigable by MERLIN.

Now students learned the necessary basic

knowledge about path planning and the basics about

the inverse kinematics of a car-like mobile robot. In

a next experiment they are able to combine this

knowledge to perform experiments related to

docking or the parking problem.

5 CONCLUSIONS AND FUTURE

WORK

According to our knowledge, we present the first

remote experiment related to the inverse kinematics

and the path planning for nonholonomic car-like

mobile robots, which is accessible via the internet.

The experiment starts from the basic motions of

nonholonomic mobile robots and demonstrates the

students how to combine these basic motions to get

some complex maneuvers. The students compare

several dedicated maneuvers for moving MERLIN

to a certain configuration. They are introduced to

sources of deviations (cf. 0, Zysko, 2004b) and the

effect of these deviations on the final configuration.

The students learn to estimate the expected

deviations when these maneuvers are performed

with the real hardware and they can choose the

appropriate maneuver for the individual needs.

In the end, our experiment provides some path

planning approaches (cf. 0) to give the students an

idea how they can use the gained knowledge about

inverse kinematics and introduces them to the

following experiment for the docking or parking

problem.

In the future we intend to extend the existing

experiments with algorithms using feedback in order

to achieve better accuracy.

ACKNOWLEDGMENT

Part of this work has been supported by the LearNet

project (http://www.learnet.de/) and a cooperation

between the University of Würzburg and the

"Virtuelle Hochschule Bayern" (

http://www.vhb.org/),

whereas LearNet was funded by the German

Ministry of Education and Research. We appreciated

the financial support provided for part of this

research within the “EU-India” program of the

European Union and the PPP China program by the

German Academic Exchange Service DAAD.

REFERENCES

Casini M., Prattichizzo D., Vicino A., Remote Control Of

A Lego Mobile Robot Through The Web, IFAC

Workshop Grenoble 2004

Schilling, K., Roth H., Roesch O. (eds.), Tele-Education in

Mechatronics Based on Virtual Laboratories, R.

Wimmer Verlag, (2001)

Schilling K., Popescu D., Meng Q., Roth H., Mobile

Roboter, D. Schmid, G. Gruhler, A. Fearns (eds.),

eLearning - Experimente und Laborübungen zur

Automatisierungstechnik über das Internet, Verlag

Europa Lehrmittel 2003, p. 137 - 146

Schilling K., Spilca C., Roth H., Levesque J.-F.,

Experiment zur PID-Regelung des Motor eines

mobilen Roboters, D. Schmid, G. Gruhler, A. Fearns

(eds.), eLearning - Experimente und Laborübungen

zur Automatisierungstechnik über das Internet, Verlag

Europa Lehrmittel 2003, p. 96 - 102.

Dormido, S. (ed.), Proceedings IFAC Workshop on

Internet Based Control Education,Pergamon Press

(2001)

Bicchi A., Caiti, A., Pallottino, L., Tonietti, G. Online

Robotic Experiments for TeleEducation at the

University of Pisa. Int. Journal of Robotic Systems, (in

press), 2004

Zysko G., Barza R., Schilling K., 2004. Tele Lab Using

Non-holonomic Car-Like Mobile Robot, IFAC

Workshop Grenoble 2004

Zysko G., Barza R., Schilling K., Lei Ma, Driewer F.,

2004. Remote Experiments on Kinematics and Control

of Mobile Robots, In Proceedings 5th IFAC

Symposium on Intelligent Autonomous Vehicles, IAV

2004, Lisbon

Latombe J.-C., Robot Motion Planning, Kluwer Academic

Publishers, 1993.

ICINCO 2005 - ROBOTICS AND AUTOMATION

434