See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/285322233

Gas Detection and Mapping Using an Autonomous Mobile Robot

Conference Paper · September 2015

CITATIONS READS
3 1,624

6 authors, including:

Mohammed Faisal Mansour Alsulaiman

Kuwait College of Science & Technology King Saud University
53 PUBLICATIONS 842 CITATIONS 155 PUBLICATIONS 4,140 CITATIONS

SEE PROFILE SEE PROFILE

Khalid Al-mutib Ramdane Hedjar

King Saud University King Saud University
78 PUBLICATIONS 731 CITATIONS 106 PUBLICATIONS 1,039 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mohammed Faisal on 01 December 2015.

The user has requested enhancement of the downloaded file.

Gas Detection and Mapping Using an Autonomous
Mobile Robot
Mohammed Faisal, Mansour Alsulaiman, Khalid Al-Muteb, Ramdane Hedjar, Hassan Mathkour, and Muhammad Emaduddin
Center Of Smart Robotics Research (CSRR), College of Computer and Information Sciences (CCIS),
King Saud University, Riyadh, Saudi Arabia P.O.Box 5117, Riyadh 11543
{mfaisal, msuliman, muteb, hedjar, mathkour, memaduddin}@ksu.edu.sa

Abstract 4 explains the experiments to test the system; and the conclusion
is given in section 5.
The field of robotics has apparently appeared to replace jobs that
humans usually dislike doing, jobs that have low tolerances for 2. Literature Review
faults or that are dangerous for humans. For these reasons, mobile
robots have been used in many applications, such as gas detection, 2.1. Mobile Robots in Hazardous Environme nts
security patrolling, and search and rescue. Nowadays, many gases
that are dangerous for humans may pollute the environment. Gases Nowadays, many jobs that are hazardous for humans are done
can leak from many causes, such as mining disasters, war, and using mobile robots. Y. Guo and Z. Qu used a mobile robot for
manufacturing disasters. In this paper, we introduce a new mobile security patrolling [1]. Their main work was to design feasible
robotics system, which uses a mobile robot for navigation within trajectories and motion control to give a robot the ability to move
hazardous environments to detect, localize, and create in a free path, covering all the points in the monitoring area [1].
concentration maps of various gases that are deemed hazardous in This enhanced previous work by focusing on the mobile robot’s
household and industrial environments. The generated gas maps coverage area and motion planning. Another study using a robot
can be used for many purposes; for example, they might be used by for security was done by Zacharie and Mbaitiga [2]. The work
humans to avoid the areas with detected gas or by a rescue team proposed a human-detection system based on a Gaussian
during rescue operations. distribution. In this system, the robot detects a human presence
Keywords: Gas Detection; Mobile Robot; Hazardous
Environment
and then sends an aural beep to warn a supervisor. This work
focused on image processing. P. Biber, S. Fleck, and T. Duckett
1. Introduction [3] used a 2D laser scanner and an omni-directional camera to
build a 3D map of an environment for a robotic security guard to
Hazardous environments present special challenges for the use. S. Soldan, J. Welle et al. used autonomous mobile robots that
accomplishment of navigational tasks. Many kinds of hazardous are equipped with several remote-gas sensing devices for the
materials are present in the industrial and general indoor detection and localization of escaped hazardous gases [4].
environments around us, including radiological materials and Finally, the Numbat robot was proposed by the Australian
toxic and harmful gases. We will focus the definition of a Commonwealth Scientific and Industrial Research Organization
hazardous environment to mean the presence of harmful gases (CSIRO) [5] for use in search and rescue tasks.
for humans within any indoor industrial, household, or
2.2. Mobile Robot Navigation
workplace setup. With the addition of each harmful or toxic
material, the problem of hazard detection and the robotic Many techniques have been used for robot navigation, including
exploration of the hazardous environment complicates itself fuzzy logic, genetic algorithms, and neural networks. A fuzzy-
many times. This paper proposes a system that uses a mobile logic based system for indoor navigation was used in [6]. The
robot to navigate hazardous environments in order to detect, authors used a Wheeled Mobile Robot (WMR), applying a fuzzy-
localize, and create concentration maps of various gases that are logic system for target-tracking control. However, their
deemed either hazardous or critical for household and industrial navigational fuzzy logic did not include obstacle avoidance.
environments. Two modules are used in performing this Another on-line robot navigation system was introduced in [7].
application: (1) a navigation module and (2) a gas-detection This on-line navigation system used two fuzzy-logic controllers,
module. The navigation module navigates the robot in the a Tracking Fuzzy-Logic Controller to navigate the robot to its
hazardous environment. The gas-detection module detects, target and an Obstacle-Avoiding Fuzzy-Logic Controller to
localizes, and creates the concentration maps of gases. The avoid obstacles. Another study used four fuzzy-logic controllers
proposed system builds concentration maps of Oxygen (O2), in a hierarchy for mobile robot navigation [8]. Three fuzzy-logic
Carbon monoxide (CO), and Methane (CH 4) gases. This paper controllers were used to navigate and to avoid an obstacle, while
is organized as follows. In section 2, a literature review is the fourth was used as a supervisory controller. Another
presented. The proposed system is explained in section 3; section navigational system using a camera and fuzzy-logic control was
introduced in [9]. Other soft computer techniques, such as 3.2. Gas-detection module
genetic algorithms and neural networks, were here used to
improve the fuzzy logic’s navigational control. A genetic The gas-detection module detects, localizes, and creates the
algorithm was used to tune the parameters of the fuzzy-logic concentration map of gases.
control in [10]. 3.2.1. Deployment of Gas Sensor
The MultiRAE Lite (Figure 2) from RAE Systems was chosen
3. Proposed System as the monitoring solution for indoor environments. This
The proposed system uses a mobile robot to navigate within diffusion-based gas sensor can detect multi-gas leakage with a
hazardous environments, detecting, localizing, and creating very short delay of a few milliseconds. The MultiRAE Lite can
concentration maps of various gases. Two modules are used in be configured to communicate with the onboard embedded PC
performing this application: (1) a navigation module and (2) a of the PowerBot (more details about the robot in the
gas-detection module. The navigation module navigates the experimental section) via an RS232 link. Various features of this
robot in the hazardous environment. The gas-detection module gas sensor include the following.
detects, localizes, and creates the gas-concentration maps.
 30 interchangeable sensor options, including PID6 for
3.1. Robot Navigation Module VOCs (Volatile Organic Compounds), an NDIR (Non-
Dispersive Infrared Sensor) for CO 2, and a catalytic
Two different fuzzy-logic controllers, the Navigation Module sensor for combustibles.
(NM) and Obstacle Avoiding Module (OAM), navigate the
 Continuous data logging (6 months for 5 sensors,
PowerBot robot through a dynamic, unknown environment. The
24x7).
NM navigates the robot to its target. It has two inputs, the angle
 Equipped to measure Oxygen, Methane, Sulfur
between the direction of the robot to the target and the x-axis
Dioxide, Carbon Monoxide, Ammonia, and Methane
(error angle), and the distance between the robot and the target.
concentrations simultaneously.
The outputs of the NM are the velocities of the left and right
motors. The NM is implemented in seven membership functions  Operating temperature of -4° to 122°F (-20° to 50°C).
for inputs and outputs. The OAM avoids static and dynamic
obstacles (if the distance between the robot and an obstacle is
less than 50 cm). The distance between the left (LD), front (FD),
and right (RD) sides of the robot and any obstacles are the inputs
of the OAM, which uses laser devices to acquire these distances.
LD is from 50 to 30 degrees, FD is from 30 to -30 degrees, and
RD is from -30 to -50 degrees relative to the front of the robot.
The outputs of the OAM module are the velocities of the left
(LV) and right (RV) motors. LV and RV in the OAM are
implemented in three membership functions. Figure 1 details
the robot navigation module.
Start

Initialization and getting the desired path

Figure 2: MultiRAE LITE Gas Sensor.

Reach the
target
Yes Stop 3.2.2. Integration of Sensor with the System
NO The Multi-RAE gas sensor is linked with the robot’s embedded
Read Laser sensors data PC using an RS-232 link. The sensor produces real-time
observations of gas concentrations, which are captured by the
No
Obstacles
Yes
gas-detection module within the robot. This module queries the
No sensor for the gas-sensor data, which is then linked to the current
Error Angle Navigation
Obstacle LD position of the robot. The gas concentration is mapped onto a
Avoiding FD
Distance Module (NM)
Module (OAM) RD grid at the particular location where the observation was taken.
The gas sensor collects data for up to three gases for each
observation. The names of each gas and their units of
WR
WL

Mobile
Platform
measurement are explained below.

Figure 1: Detailed flowchart of robot navigation module.

- Oxygen (O2)
Oxygen is present almost everywhere in a well-ventilated indoor literature regarding the creation of hazard maps for harmful
environment. The gas maintained exactly 20.9% composition materials and radiation [11].
among all the gases present in our experimental environment. The researcher measured the performance of the survey
We measured oxygen in terms of its percentage composition of techniques using certain parameters.
all gases within the environment.

- Methane (CH4)
We produced trace quantities of Methane via our gas source
during the experiment. Only trace quantities were used in order
to avoid any ignition. We measured methane in terms of the
percentage of the Lower Explosive Limit (LEL) of this gas.
Mixtures of dispersed combustible materials (such as gaseous or
vaporized fuels and some dusts) and air will burn only if the fuel
concentration lies above a well-defined bound determined
experimentally, termed Explosive Limits. This means that the
chance of ignition is only likely, by definition, when Methane
%LEL crosses over the 100% mark. During all our experiments,
Figure 3: PowerBot installed with MultiRAE Lite gas sensor.
Methane concentration never exceeded 4% LEL. Methane was
measured via NDIR, with resolution of 1% LEL. The gas-mapping parameter (1) “Average Time Taken to
Complete the Survey” shows the total time consumed by the
- Carbon Monoxide (CO) survey process. This parameter can be crucial for applications
We produced significant quantities of Carbon Monoxide during where the time needed to assess harmful gas distributions is
our experiments. Carbon Monoxide was measured in parts per critical. The gas-mapping parameter (2) “Average Distance
million (ppm) in our experiments. As a relative measurement, from Gas-Source at the Time of First Detection” is important
ppm is not affected by pressure. This measurement is derived mainly because detection at the farthest distance from a
from the units of milligrams per cubic meter. The conversion hazardous gas source can save both equipment and lives when
formula is: searching for harmful gases. The gas-mapping parameter (3)
𝑚𝑔 0.08205.𝑇 “Average of X,Y Locations of Points where Gas was Detected”
𝑝𝑝𝑚𝑣 = 𝑚3 . 𝑀 (1) is important since this indicates the direction of the gas source
where mg/m is milligrams of pollutant per cubic meter of air in
3
relative to the robot and the rescue or disaster-management
the atmospheric pressure at sea level, T is the ambient crew.
temperature in Kelvins, and M is the molecular mass of the air The location of the gas source was kept constant throughout all
pollutant. experiments using different survey techniques. These
During our experiments, the sensor detected a maximum of 8 experiments focused on Carbon Monoxide using different
ppm of Carbon Monoxide, indicating a relatively low quantity survey scenarios. In addition, we mapped the concentrations
of gas concentration. The United States Occupational Safety and when detecting two gases at the same time, namely Methane and
Health Administration prohibits exposure to over 50 ppm for Carbon Monoxide. The mapped results have Cartesian x and y
longer than 8 hours a day. Even brief exposures to quantities coordinates attached to observed gas concentrations.
higher than 100 ppm can cause significant health damage to
humans. 4.1. Experimental Results
4. Experiments Gas-mapping results for Carbon Monoxide (CO) are shown in
The mobile robot used in this work was a PowerBot (Figure 3), Figure 4–Figure 7 for various surveying techniques. For each,
a wheeled mobile robot from Adept Mobile Robots. the gas source was at (0 m, -2.5 m). Most results indicated the
It is an automated, differential-drive, guided vehicle specially presence of CO in the vicinity of the source, but the gas spread
designed and equipped for autonomous operation, intelligent far and wide, and a significant number of outlier readings may
delivery, and handling of large payloads. The proposed system be seen. Figure 8 shows a comparison of gas-mapping between
was able to detect and map concentrations of various gases that CO and CH4 for the same scenario.A marked difference in
are either deemed hazardous or critical for household and Figure 8 detected gas concentrations is evident in the results. A
industrial environments. The goal of the experimental strategy video recording of the experiments is available
was to generate maps of gases. Various survey techniques were on the Youtube Channel at https://www.youtube.com/channel/
used to map the O 2, CO, and CH4 concentrations within an UCn_bdbXUFV1NyCpl2oXUvuw. T ABLE 1 compares the
indoor environment. These techniques were derived from the performance of the survey techniques using the parameters
defined before. Bolded numbers indicate the best result for each was between 3 and 4, while the lowest (far from the gas source)
parameters.In terms of early detection, values for Average was less than 1.
Distance from Gas-Source at the Time of First values suggest
that more-or-less all survey techniques detected gases at similar
distances. The cloverleaf surveying technique, however, stands
out as the most suitable for estimating the direction of an
incoming flow of gas.

Table 1: Gas-Mapping Performance Parameters

Survey Average Average Locations of

Technique Time Taken Distance first
to Complete from Gas- detection of
the Survey Source at the gas
(std. dev) Time of First
Detection
Lane 254 secs (5.17) 2.66 m (-0.02,-0.58)
Zigzag 264 secs (5.11) 2.64 m (1.31,-0.56)
Cloverleaf 566 secs (8.22) 2.61 m (0.23.-0.95)
Star 272 secs (5.92) 2.52 m (-0.84.-0.68)

The Lane technique completed the survey within the minimum

time (254 sec). However, as can be seen in Figure 4, it did not
detect a high concentration of the gas; most of its detection was
below 0.5. Figure 5: Gas concentration map along with respective robot path trajectory
using the Zig- Zag survey technique.

The Cloverleaf technique completed the survey in 266 sec. As

can be seen in Figure 6, it detected three different levels of
concentration, the highest concentration (near to the gas source)
in the range 1–1.5 and the lowest (far from the gas source) less
than 0.5.

Figure 4: Gas concentration map along with respective robot path trajectory
using the Lane survey technique.

The Zigzag technique completed the survey in 264 sec. As can

be seen in Figure 5, it detected four different levels of Figure 6: Gas concentration map along with respective robot path trajectory
concentration, the highest concentration (near to the gas source) using the Cloverleaf survey technique.
The Star survey technique completed the survey in 272 sec. As 5. Conclusion
can be seen in Figure 7, it detected three different levels of
concentration, the highest concentration (near to the gas source) In this paper, we used a PowerBot mobile robot to navigate in a
in the range 1–1.5 and the lowest (far from the gas source) less hazardous environment, detecting, localizing, and creating
than 0.5. concentration maps of various hazardous gases. The generated
maps could be used by humans to avoid the detected gas areas,
by a rescue team during rescue operations, or in many other
applications. The experimental results of the proposed system
illustrated its successful operation. It built concentration maps
in environments with many levels of concentration. We also
compared various survey techniques. Choosing the best
technique depends on the most important performance
parameter for a particular application.

Acknowledgments

This work was supported by Center of Smart Robotics Research

(CSRR),College of Computer and Information Sciences (CCIS),
King Saud University. The authors are grateful for this support.

REFERENCES
[1] Y. Guo and Z. Qu, "Coverage control for a mobile
robot patrolling a dynamic and uncertain
environment," in Intelligent Control and Automation,
Figure 7: Gas concentration map along with respective robot path trajectory
2004. WCICA 2004. Fifth World Congress on, 2004,
using the Star survey technique.
pp. 4899-4903.
The concentrations of Methane and Carbon Monoxide gases [2] M. Zacharie, "Security Guard Robot Detecting Human
taken in the same survey are illustrated in Figure 8. As can be Using Gaussian Distribution Histogram Method,"
seen, Carbon Monoxide gas was present in the environment in Journal of Computer Science, vol. 6, p. 1144, 2010.
higher concentration than was Methane. [3] P. Biber, S. Fleck, and T. Duckett, "3d modeling of
indoor environments for a robotic security guard," in
Computer Vision and Pattern Recognition-Workshops,
2005. CVPR Workshops. IEEE Computer Society
Conference on, 2005, pp. 124-124.
[4] S. Soldan, J. Welle, T. Barz, A. Kroll, and D. Schulz,
"Towards autonomous robotic systems for remote gas
leak detection and localization in industrial
environments," in Field and Service Robotics, 2014,
pp. 233-247.
[5] D. Hainsworth, "Teleoperation user interfaces for
mining robotics," Autonomous Robots, vol. 11, pp. 19-
28, 2001.
[6] R. Rashid, I. Elamvazuthi, M. Begam, and M. Arrofiq,
"Differential drive wheeled mobile robot (WMR)
control using fuzzy logic techniques," in
Mathematical/Analytical Modelling and Computer
Simulation (AMS), 2010 Fourth Asia International
Conference on, 2010, pp. 51-55.
[7] M. Faisal, R. Hedjar, M. Al Sulaiman, and K. Al-
Mutib, "Fuzzy Logic Navigation and Obstacle
Figure 8:Methane and Carbon Monoxide simultan eous concentration mapping Avoidance by a Mobile Robot in an Unknown
for the Zig- zag survey technique. Dynamic Environment," INTERNATIONAL
 JOURNAL OF ADVANCED ROBOTIC SYSTEMS,
vol. 10, 2013.
[8] F. Cupertino, V. Giordano, D. Naso, and L. Delfine,
"Fuzzy control of a mobile robot," Robotics &
Automation Magazine, IEEE, vol. 13, pp. 74-81, 2006.
[9] V. Raudonis and R. Maskeliunas, "Trajectory based
fuzzy controller for indoor navigation," in
Computational Intelligence and Informatics (CINTI),
2011 IEEE 12th International Symposium on, 2011, pp.
69-72.
[10] G. Narvydas, R. Simutis, and V. Raudonis,
"Autonomous mobile robot control using fuzzy logic
and genetic algorithm," in Intelligent Data Acquisition
and Advanced Computing Systems: Technology and
Applications, 2007. IDAACS 2007. 4th IEEE
Workshop on, 2007, pp. 460-464.
[11] L. A. Hansen, "An Evaluation of a Networked
Radiation Detection System," DTIC Document2010.

View publication stats