A Novel Wall Following Algorithm For Mobile Robots

Min Raj Nepali, Nishesh Yadav, Dubey Aditya Housila Prasad, Susheel Balasubramaniam, Venkatesh EN &
Ashutosh
International Journal of Robotics and Automation (IJRA), Volume (5) : Issue (2) : 2014 15
A Novel Wall Following Algorithm For Mobile Robots
Min Raj Nepali nminraj92@gmail.com
Research Associate
Nitte Meenakshi Institute of Technology
Bangalore, 560064,India
Nishesh Yadav hiddenshadowy@gmail.com
Final Year Bachelors,EEE
Dubey Aditya Housila Prasad adityad031@gmail.com
Research Associate
Susheel Balasubramaniam susheels88@gmail.com
Research Associate
Venkatesh EN venkateshen1@gmail.com
Research Associate
Ashutosh ashu.studsat@gmail.com
Research Associate
Abstract
This paper presents how a mobile robot can be directed to follow a wall using 2D data obtained
with the help of a laser range finder. The robot is controlled using a simple kinematic controller
and the controller is embedded in NXP LPC1768 microcontroller. The controller guides the robot
to follow the wall along the required direction while maintaining a constant distance from the wall.
The data obtained from the laser range finder is analyzed and processed to generate signals to
control the speed as well as orientation of the robot. Implementation results show that the
proposed algorithm controls the robot to follow walls, detecting the edges and corners
successfully.
Keywords: Kinematic Controller, 2D Laser Range Finder, Autonomous Mobile Robot, Edge
Detection, Corner Detection, wall Following.
1. INTRODUCTION
Autonomy has become a key concept in mobile robot navigation to complete desired task
efficiently and with least human intuition. Mobile robots have numerous applications in various
fields like space exploration, defence, education, etc. Considering different applications the robots
should be incorporated with various algorithms like path planning, obstacle avoidance, etc. Wall

Ashutosh
following is one of those algorithms which is to be implemented to make the robot autonomous. In
this experiment laser range finder was used to map the environment and later on that information
was used to navigate the robot and follow the wall accordingly. The perceived environment was
classified as edges and corners based on different features and the robot was duly driven.
Extensive research has been performed in the area of mobile robots and their applications.
A wall following method presented in [1] [2] uses IR sensors and divides the environment into a
number of cells representing the wall as set of lines, points, etc. Another wall following algorithm
presented in [3] [4] uses image processing to extract features like edges and corners after
capturing the images or videos through a camera by using background subtraction method which
later on is fed to control systems to drive the robot. Visual servoing was implemented to track and
follow the wall using camera. In [5] [6] SONAR is used to perceive the environment and a Fuzzy
logic controller is implemented to control the robot. Fuzzy logic controllers prove to be one of the
best controllers since they represent real world knowledge and handle real world uncertainties so
adeptly[7] [8].The paper [9] proposes a robust model to avoid obstacles which is further optimized
to achieve wall following task by using a suitable neural controller. The sensors used to obtain
information from the environment were sonar sensors.
The three important questions that have to be countered by the robot are (Where I am, Where to
go, and how to go). To answer these questions correctly the robot has to have precise knowledge
regarding the environment. The robot then should adjust itself relatively to the environment.
2. ROBOT MODEL
FIGURE 1: Showing the relationship between Global frame and Robot body Frame.
Throughout the experiment the robot was considered a rigid body on wheels moving on a plane
surface. In order to postulate the position of the robot on the plane we create a relation between
the global reference frame of the plane and the local reference frame of the robot, as in figure
above. The axes X and Y define a random inertial basis on the plane as the global reference
frame or the origin. To specify the position of the robot, we choose a point O on the robot chassis
as its position reference point. The basis {x, y} defines two axes relative to O on the robot chassis
and is thus the robot’s local reference frame. The position of the robot O in the global reference

Ashutosh
frame is specified by coordinates x and y, and the angular variation between the global and local
reference frames is given by ‘α’. We can describe the pose of the robot as a vector with these
three elements.
XY
x
O y

 
   
  
To describe robot motion in terms of component motions, it will be compulsory to map motion
along the axes of the global reference frame to motion along the axes of the robot’s local
reference frame. Of course, the mapping is a function of the current pose of the robot. This
mapping is achieved by orthogonal rotation matrix:
cos sin 0
( ) sin cos 0
0 0 1
R
 
  
 
   
  
This matrix can be used to map motion in the global reference frame {x, y} to motion in terms of
the local reference frame{x, y}.

Ashutosh
3. DEVELOPMENT OF THE ALGORITHM
START
SET MINIMUM DISTANCE
TO BE MAINTAINED FROM
THE WALL
START SCANING
CHECK THE
MINIMUM
DISTANCE
KEEP FOLLOWING THE
WALL
GENERATE VELOCITY FOR
LEFT AND RIGHT WHEEL
TO MAINTAIN THE
MINIMUM DISTANCE
EDGE OF THE
WALL
DETECTED
FOLLOW THE WALL
CORNER
DETECTED
TURN AROUND THE
CORNER
Y
Y N
Y
N
N
FIGURE 2: Flow diagram of the controller.
The model assumes there is always a wall present somewhere in the range of the laser range
scanner. The wall is perceived as a 2D plane. The wall detected is classified as edges and

Ashutosh
corners based on different features. Segmentation, split and merge techniques may also be
employed to classify the wall or to extract different features related to the wall. As far as the edge
is considered the controller generates linear velocity but as soon as a corner is detected the
controller generates angular velocity such that the robot is just able to tackle the corner. The laser
is mounted in the robot such that the axes of the robot and the axes of the laser coincide. The
laser scanner provides us with the distance between the wall and the reference point. Here the
reference point is considered as the point where the laser range finder is mounted. The distance
obtained from the laser range finder is resolved in its respective x component, y component and
an angle from which exact location of the robot is obtained.
The object O mounted on the top of the robot R is the laser range finder. It is clearly visible that both the
axes of robot and the laser are same. As shown in the figure the point P which is part of the wall is being
resolved into PS and PQ as y and x component respectively. The robot will always try to maintain minimum
distance with the wall and always try to orient PS perpendicular to the wall.
4. CONTROLLER IMPLEMENTATION
The laser range scanner scans the environment. The scanner has capability to scan a range of 270 degrees
that is from -135 to +135 as shown in the figure. The angular resolution is set to 0.5, which implies that in
one scan time the laser outputs a bulk of 541 data. In this experiment two zones of 45 degrees were
considered that is from -135 to -90 degrees and 135 to 90 degrees.
FIGURE 3: Robot position with respect to wall.

Ashutosh
.
FIGURE 4: Scan range of LMS 100.
The controller is designed in such a way that it particularly takes two cases into consideration. If
the wall is present on the left side of the robot, the scanned data corresponding from 135 to 90 is
taken into consideration for controlling the motion of the robot. If the wall is present on the right
side of the robot the scanned data corresponding from -135 to -90 is taken into consideration for
controlling the motion of the robot. The controller implemented here is simple kinematic controller,
which is capable of generating sufficient angular and linear velocities to drive the robot in a proper
way. A threshold distance (dth ) is defined for the robot to maintain while following the wall. If
distance between the robot and wall is greater than the threshold distance the robot is driven
towards the wall and vice-versa. The controller may be described as follows:
Where,
V=linear velocity generated
W=angular velocity generated
K_V=velocity error gain
K_d=distance error gain
K_t=theta error gain
dperp =perpendicular or minimum distance to be maintained while following the wall
derror =error in distance
aerror =error in angle
5. EXPERIMENTAL SETUP AND RESULTS
A series of experiments were conducted to ensure the controller works precisely. The gain values were
determined by trial and error and the whole experiment was carried out in an indoor environment. The
mobile robot was made to follow a series of walls and the performance of the controller was observed. The
result shows that the laser is being able to detect the edges and corners successfully. After detecting the
edges and corners respective control signals are being generated with the help of which the controller is
driving the robot by passing down the control signals to the motors. The controller was experimentally shown
to work and the series of walls were successfully followed.
V W 

Ashutosh
FIGURE 5: Model of LMS 100.
(a) (b)
(c) (d)
(e) (f)
(g) (h)
(i) (j)
FIGURE 6: Experimental results (a) to (j) depicting the detection of edge and corner and maneuvering the
robot accordingly.

Ashutosh
6. CONCLUSION AND FUTURE SCOPE
Mobile robots are capable to move around in their surrounding location and are not fixed to one location
when tasks like navigation, survey, etc are considered. An autonomous robot knows some facts about
where it is and how to reach various goals and or waypoints lying on its way. "Localization" or knowledge of
its current position is obtained with the help of one or more means, using sensors such motor encoders,
vision, lasers and global positioning systems. A novel wall following algorithm for mobile robots equipped
with a 2D laser range finder has been presented. In this implementation, two cases i.e., if the left wall is
present and if the right wall is present were taken into consideration. In scenarios such as if a sharp corner
is present right in front of the robot; further expansion to the algorithm is required. However, the mobile robot
was experimentally shown to reliably follow a series of walls given the constraints mentioned above. In an
unknown environment it is easy for a robot to navigate, if it can find wall like structures and follow it. Wall
following method would better help robot rather than path following if structures like wall are present in the
unknown environment. It happens in many cases like underground, underwater, etc wall like structures are
present and by making use of them wall following robot can easily navigate through the environment due to
which the robot gains more autonomy. This kind of autonomous algorithms can be incorporated with SLAM.
It would be easier to carry out the SLAM if the robot be totally autonomous. Wall follow algorithm may
perform extremely well if it is carried along with mapping. From mapping we can easily obtain edges and
corners of the wall and make robot to follow accordingly.
7. REFERENCES
[1] P. Van Turennout, G. Honder d, L. J. vanSchelven, Wall-following control of a Mobile Robot,
IEEE, International Confer ence onRobotics and Automation, pp. 280-285, May,1992.
[2] J. Borenstein, Y. Kor en, The Vector Field Histogram – Fast Obstacle Avoidance for Mobile
Robots, IEEE Transaction on Robotics and Automation, Vol. 7, No. 3, pp. 278-288, June,1991.
[3] R. Frizera Vassallo, H. J. Schneebeli, J. SantosVictor, Visual navigation: combining visual
servoing and appearance based methods,Proceedings of the International Symposium
onIntelligent Robotic Systems, SIRS”98,Edinburgh, Scotland, July, 1998.
[4] Francisco Bonin-Font, Alberto Ortiz, Gabriel Oliver, Visual Navigation for MobileRobots: A Survey.
2008.
[5] R. Carelli and E. Freire, “Stable Corridor Navigation Controller for Sonar-Based Mobile Robots,”
Technical Report, INAW, Universidad Nacional de San Juan, San Juan, 2001.
[6] D. Ratner and P. McKerrow,“Navigation an Outdoor Robot along Continuous Landmarks with
Ultrasonic Sen- sing,” Robotics and Autonomous Systems, Vol. 45, No. 1, 2003, pp. 73-82.
doi:10.1016/S0921-8890(03)00096-4.
[7] R. Malhotra and A. Sarkar, “Development of a Fuzzy Logic Based Mobile Robot for Dynamic Obstacle
Avoidance and Goal Acquisition in an Unstructured Enviroment,” IEEE/ASME Proceeding of
International Conference on Advanced Intelligent Mechatronics, Kobe, 20- 24 July 2003, pp. 235-247.
[8] P. M. Peri, “Fuzzy Logic Controller for an Autonomous Mobile Robot,” Master Thesis, Cleveland State
University, Cleveland, 2005.
[9] R. Braunsting, J. Mujika and J. P. Uribe, “A Wall Fol- lowing Robot with a Fuzzy Logic Controller
Optimized by a Genetic Algorithm,” Fuzzy Systems, Vol. 5, 1995, pp. 77-82.
[10] M. Ergezer, “Multivariable Control Method for Wall Tracking Robot,” Project Paper,Department of
Electrical and Computer Engineering, Cleveland State University, Cleveland, 2006.
[11] R. Benporad, M. Di Marco and A. Tesi, “Wall-Following Controller for Sonar-Based Mobile Robots,”
Proceeding of IEEE Conference on Decision and Control, San Diego, 10-12 December 1997, pp.
3063-3068.
[12] Specifications and Operating Commands for SICK LMS100 Laser Range Finder. Can be found on
www.sick-automation.ru/images/File/pdf/DIV05/LMS100_manual.pdf.

Ashutosh
[13] R. Clark, A. El-Oser y, K. Wedeward and S.Bruder, A Navigation and Obstacle Avoidance
Algorithm for Mobile Robots Operating inUnknown, Maze-Type Environments, Proc.International
Test and Evaluation AssociationWorkshop on Modeling and Simulation, LasCruces, NM,
December 2004.

A Novel Wall Following Algorithm For Mobile Robots

More Related Content

What's hot (19)

Similar to A Novel Wall Following Algorithm For Mobile Robots (20)

More from Waqas Tariq (20)

Recently uploaded (20)

A Novel Wall Following Algorithm For Mobile Robots