Abstract The main purpose of this research is to design an autonomous robot that can navigate around an indoor environment to provide information about the particulate matter 2.5 (PM¬2.5) concentration on the said environment. Also, the mobile robot is equipped with an air filtering device that would only operate whenever there is high concentricity of the particulate matter. The capability of the robot to mobile robot to navigate around and filter the air intelligently would decrease the battery consumption, which is one of the problem of intelligent robots. Also, the robot would provide a map of the concentricity of particulate matter inside the indoor environment which would give the operators about ideas where are the ideal places or the hazardous places where there is high concentricity of PM2.5.
The map of the concentricity of PM2.5 will be displayed on the remote computer that controls the operation of the autonomous mobile robot automatically. One of the main objective of this paper is to make the system fully autonomous and remove the requirement of having a human operator to avoid the hazardous effects of the particulate matter to humans. Acknowledgements First of all, I would like express my very great appreciation to my advisor, Professor Chian-Song Chiu, who made all of these works possible, without him I would not be able to grasp even the smallest knowledge to start these works. I would also like to offer my special thanks to Dr.
Angelo Beltran, Jr., my professor in Adamson University who helped me to become a student here in Chung Yuan Christian University. I would also extend my thanks to the staffs of the Electrical Engineering Department of CYCU for their help in offering me resources in the course of this program. I would also like to express my gratitude to my laboratory mates and friends, ???, ???, ???, ???, ???, ???, ???, ??, ???, who guided me when I first came in Modern Control and Applications Laboratory. I am particularly grateful for the assistance given by??? and???.
To my father, Carlito Santiago, who gave me moral and financial support and was there all the time through high and lows while I was taking up my Master Degree. And to my beloved mother, Lucita Santiago, who, although no longer with us, continues to inspire me through her examples and dedication to academe. My inspiration to finish these works through sleepless nights is to make you proud up there. And lastly, I would like to thank God Almighty for giving me the strength, knowledge, perseverance and opportunity to undertake this research study and complete it satisfactorily. Without His great blessings, this achievement would not even be possible. Table of Contents ?? i Abstract ii Acknowlegements iii Table of Contents iv List of Figures viii List of Tables xviii Chapter 1 Introduction 1 1.1 Background of the Study 1 1.2 Research Motivation and Literature Review 3 1.2.1 Autonomous Mobile Robot Advancements……………………………………..……3 1.2.2 Particulate Matter 2.5 (PM2.5)………………………………………………6 1.3 Organization of Dissertation 7 Chapter 2 Autonomous Mobile Robot 9 2.1 Kinematics of a Mobile Robot 10 2.2 Sonar Sensor 12 2.2.1 Sonar Sensor Data Inaccuracies……………………………………………….16 Chapter 3 Introduction to Indoor Positioning System and PM2.5 Sensor 18 3.1 StarGazer Indoor Positioning System 20 3.1.1 Operation of Indoor Positioning System 21 3.1.2 Map Building Mode 23 3.1.3 Data Acquisition 23 3.2 Particulate Matter 2.5 (PM2.5) Sensor 25 3.2.1 Microcontroller (Arduino Uno) 27 Chapter 4 Methodology 31 4.1 Interval Type-2 Fuzzy-PID Dual-Mode Controller 31 4.2 Interval Type-2 Fuzzy Logic Controller 32 4.3 Type-2 Fuzzy Logic System 35 4.4 Proportional Integral Derivative Controller (PID Controller) 38 4.5 Type-1 Fuzzy Logic Controller 40 4.5.1 Fuzzification…………………………………………………………………………………..41 4.5.2 Fuzzy Rules ……………………………………………………………………………….43 4.5.2 Defuzzification …………………………………………………………………………..45 Chapter 5 Automated Air Filtering and PM2.5 Real-Time Mapping System…………58 Chapter 6 Experiment and Results 63 6.1 Type-1 Fuzzy Controller Simulation Results 65 6.2 PID Controller Simulation Results 69 6.2 IT2FPIDDMC Controller Simulation Results 73 Chapter 7 Conclusion and Future Works 78 References 79 List of Figures Figure 2-1: P3DX mobile robot’s physical dimension and sonar sensor orientation 9 Figure 2-2: Pioneer 3-DX Mobile Robot……………………… 10 Figure 2-3: A model of a two-wheeled mobile non-holonomic robot …………… 11 Figure 2-4: Principle of an active sonar 13 Figure 2-5: Acquired data from sonar sensors 14 Figure 2-6: Firing rate of sonar sensor 15 Figure 2-7: Specular Reflection Effect 17 Figure 3-1: Hagisonic StarGazer Indoor Positioning Device 18 Figure 3-2: Passive Landmark attached to the ceiling 20 Figure 3-3: Actual setup of the StarGazer Indoor Positioning Device 20 Figure 3-4: StarGazerMonitor UI.
The computer serial setting is boxed for setting up the correct computer settings 21 Figure 3-5: The data is acquired by the StarGazerMonitor 21 Figure 3-6: Map building mode – the landmarks were already detected and displayed in the StarGazerMonitor 22 Figure 3-7: Map and data acquired from MATLAB 23 Figure 3-8: Experimental scenario – minimum light 23 Figure 3-9: Experimental scenario – light is abundant 24 Figure 3-10: Data acquired when light is abundant ……………. 24 Figure 3-11: Sharp GP2Y10 Dust Sensor 25 Figure 3-12: Sharp GPY10 Dust Sensor Pin Assignments …………………. 26 Figure 3-13: Sharp GPY10 Dust Sensor Pin Configuration 26 Figure 3-14: Experiment no. 1 – no object is inserted in dust detection area 28 Figure 3-15: Experiment no. 2 – an object is inserted in the dust detection area 28 Figure 3-16: Initial data from the dust sensor using Arduino Uno from experiment 1 29 Figure 4-1: The schematic diagram of the IT2FPIDDMC 31 Figure 4-2: Membership function for (4) 32 Figure 4-3: Membership function for (5) 33 Figure 4-4: Key components of FLC system 33 Figure 4-5: Membership function for (6) 34 Figure 4-6: Membership function for (7) 34 Figure 4-7: PID controller system architecture 40 Figure 4-8: System architecture of the type-1 fuzzy logic system 41 Figure 4-9: Membership Function for Sonar012, Sonar34, Sonar567 43 Figure 4-10: Center of gravity defuzzification method 46 Figure 4-11: Center of sets defuzzification method 47 Figure 5-1: Mobile robot with air-filtering and PM2.5 mapping system…………….…59 Figure 5-2: Map acquired from the first run of mobile robot…………………………60 Figure 5-3: Map acquired from the second run of mobile robot…………….………..61 Figure 5-4: Data acquired from the whole system……………………………………62 Figure 6-1: Experiment scenario 1 – mobile robot following the inside contours of a structured wall 48 Figure 6-2: Experimental scenario 2 – mobile robot following the inside contours of an unstructured wall 49 Figure 6-3: Experimental scenario 3 – mobile robot following the outside contours of a structured wall 49 Figure 6-4: Experiment scenario 4 – mobile robot following the outside contours of an unstructured wall 50 Figure 6-5: Experimental scenario 1 type-1 fuzzy logic controller trajectory 51 Figure 6-6: Experimental scenario 2 type-1 fuzzy logic controller trajectory 51 Figure 6-7: Experimental scenario 1 step response curve using type-1 fuzzy controller 52 Figure 6-8: Experimental scenario 2 step response curve using type-1 fuzzy controller 52 Figure 6-9: Experimental scenario 3 step response curve using type-1 fuzzy controller 52 Figure 6-10: Experimental scenario 4 step response curve using type-1 fuzzy controller 53 Figure 6-11: Left and right wheel running velocity step response curve of type-1 fuzzy controller 54 Figure 6-12: Experimental scenario 3 PID controller trajectory 55 Figure 6-13: Experimental scenario 1 step response curve using type-1 fuzzy PID controller 55 Figure 6-14: Experimental scenario 2 step response curve using type-1 fuzzy PID controller 56 Figure 6-15: Experimental scenario 3 step response curve using type-1 fuzzy PID controller 56 Figure 6-16: Experimental scenario 4 step response curve using type-1 fuzzy PID controller 57 Figure 6-17: Left and right wheel velocity step response for Type-1 Fuzzy-PID controller 58 Figure 6-18: Interval type-2 fuzzy PID dual-mode controller trajectory 59 Figure 6-19: Experimental scenario 1 step response curve using interval type-2 fuzzy PID dual-mode controller 59 Figure 6-20: Experimental scenario 3 step response curve using interval type-2 fuzzy PID dual-mode controller 60 Figure 6-21: Experimental scenario 3 step response curve using interval type-2 fuzzy PID dual-mode controller 60 Figure 6-22: Experimental scenario 4 step response curve using interval type-2 fuzzy PID dual-mode controller 61 Figure 6-23: Left and right wheel velocity step response for interval type-2 fuzzy PID dual-mode controller 61 List of Tables Table 2-1: Pioneer 3-DX Mobile Robot Specifications 13 Table 3-1: Hagisonic StarGazer Specifications 19 Table 3-2: PM2.5 Sensor Pin Definition 27 Table 4-1: Fuzzy Inference Rule Base of Running Velocity 35 Table 4-2: Fuzzy Controller Fuzzy Rules 44 Table 5-1: Data Acquired from StarGazer and PM¬2.5 Sensor 44 Chapter 1 Introduction 1.1. Background of the Study Recent studies show evidence on the association of particulate matter (PM2.5) and other respiratory system related diseases like peripheral artery disease (PAD) 1. In previous years, several research efforts were conducted to identify the PM2.5’s concentration in several cities. During 2006–2009, increasing levels of PM2.5 concentration were observed in the 3 major cities of Taiwan. PM2.5concentration levels during different seasons showed similar fluctuations between the 3 major cities.
Winter season between December and January of the following year yielded the highest concentration levels, and summer between the months of June to August yielded the lowest concentration. Taipei and Taichung were observed to have higher concentration levels during sandstorms occurring in the month of March 2. Therefore, among several environmental issues, PM2.5 has received some considerable attention from the government and academic units. Although several types of research about PM2.5’s effect on health for decades, the Air Quality Annual Report of Taiwan published by the Environment Protection Administration Executive Yuan (EPA) did not address the pollutants of PM2.5 until the year 2013. It is also the first official publication that has included this major pollutant identified by the International Agency for Research on Cancer (IARC).
However, the information which was disclosed lately was so vague that people cannot fully realize several environmental issues and health risk that they are facing 3. According to the 2013 report, the percentage of days that exceeds air pollution quality standards among recorded days Pollutant Standards Index (PSI) was 1.42%. The value seems pretty low and harmless; however, the result did not include the PM2.5 in the monitoring report. This leads to the misconception among local residents who think the air quality is better than what they actually experience, especially for people who are living nearby the industrial zones in Kaohsiung and suffering from stinky air three or four days a week.
With these implications, several research efforts were conducted by the government and academic units in improving environmental and safety standards that involves issues about air pollutants. It has always been the goal pursued by all government units, to improve the life quality of its people. Studies were conducted to further identify its impacts and map the concentrations of air pollutants in several cities in Taiwan 4. In this paper, the concentration of PM2.5 is identified in an indoor environment. The air pollutant’s concentration is mapped inside a building to avoid these places with high concentration of the said pollutant. Additionally, as a solution to this problem, a mobile robot equipped with air filtering system navigates around the environment to simultaneously map the concentration of PM2.5 in the air and at the same time clean it.
The mobile robot is fully autonomous; therefore, an operator is not needed. It has multiple built-in sensors and computer to enable it to implement automatic navigation. A remote computer and onboard computer is included in the system to control the mobile robot and map the concentration of air pollutants, respectively. For the whole system to become fully autonomous, several studies and experiments have been conducted to improve the Autonomous Guided Vehicle (AGV) system of the mobile robot. 1.2. Research Motivation and Literature Review This section is subdivided into two parts whereas the previous studies regarding the mobile robot and PM2.5 are reviewed.
The research motivation behind this paper is also discussed in the latter part. 1.2.1. Autonomous Mobile Robot Advancements In robot development field, mobile robot applications have received some considerable attention from academic units in the past years. Autonomous mobile robots have been involved in some very complicated tasks like rescue missions, transportation of medical equipment, or surveying and mapping unknown environments 5 6 7. Navigating real-time in real-world, unknown environments is the most basic ability that an autonomous mobile robot must possess in order to do more complex tasks effectively; In order for the mobile robot to implement this, it must be equipped with several sensors. There are a wide variety of sensors to choose from in order to implement this capability in a mobile robot, a laser rangefinder is regarded as one of the most efficient and accurate sensors used in autonomous navigation but is quite expensive for just implementing a simple obstacle avoidance task.
Infrared sensors, which are cheaper than laser range finders can also be used for autonomous navigation but is quite unstable and the measured distance shows a large deviation in the actual distance 8. Also, regardless of what sensory device is used, there will be accumulated noise with the data acquired. In order to minimize these system errors, caused by either the outside noises or just the system error itself, several controllers have been previously designed to give a good control output that would meet our requirements. Traditional Proportional-Integral-Derivative (PID) controllers have been used for these purposes previously 9 10. But standalone PID controllers were proven to be ineffective in dealing with non-linear, unstructured and changing environments, especially when the environment is previously unknown. The choice of adequate methods to model and handle such uncertainties in unknown environments is crucial for robot navigation.
The fuzzy logic controller (FLC) is credited with a high degree of reliability for controlling and solving complicated systems like the AGV system. Multiple studies have also been conducted to use fuzzy systems to solve certain problems 11-14. Therefore, a lot of investigation has focused on implementing fuzzy logic control in AGV systems, in 15, Thongchai and Kawamura, designed a fuzzy controller that provides a mechanism for solving sensor data from all sonar sensors which present different information. And results showed that a complicated system can be solved, each behavior works correctly, and an emergency behavior has the highest priority to stop the robot if the obstacles are closer than the emergency distance.
In 16, the mobile robot uses a fuzzy logic controller and local navigation strategy. The basis for local navigation is provided by the concept of general perception, which passes perceptual information of the sensors on to the fuzzy system without modeling walls or obstacles. Thus, no representation of the environment is needed. In 5, a car model is used to design a fuzzy controller for obstacle avoidance for a moving vehicle.
To date, almost all the FLC implementations in robot control are based on type-1 fuzzy logic controller 17-22. The most common way is to construct the FLC by eliciting the fuzzy rules and the membership functions based on expert knowledge or through observation the actions of a human operator controlling the mobile robot. However, type-1 FLCs (designed using human experience or using learning mechanisms) have a common problem that they cannot fully handle or accommodate for the linguistic and numerical uncertainties with changing and dynamic unstructured environments as they use only type-1 fuzzy sets to represent a system 23. Type-1 fuzzy sets handle the uncertainties associated with the FLC inputs and outputs by using precise and crisp membership functions that the user believes to capture uncertainties 23. All uncertainties disappear totally because type-1 membership functions are precise 24.
This may cause problems in determining precise antecedents of linguistic and numerical uncertainties with changing unstructured environments and sensor data inaccuracies. Moreover, the design of the type-1 fuzzy sets can be suboptimal under specific robot conditions, like sonar sensors giving inaccurate data, the chosen type-1 fuzzy sets might not be appropriate anymore. This can degrade the performance of the FLC and might end up redesigning the whole system 25. A type-2 fuzzy set is characterized by a fuzzy membership function where the fuzzy sets are defined by type-1 membership functions, unlike type-1 fuzzy sets which are defined by a crisp number between 0,1.
Type-2 fuzzy sets’ membership function is three dimensional and it includes a footprint of uncertainty which provides additional degrees of freedom that makes it possible to directly model and handle uncertainties. Type-2 fuzzy is ideal to use in cases for mobile robots in unstructured environments where it is difficult to determine the exact and precise membership functions. In this paper, the type-2 fuzzy logic system has been implemented to the mobile robot optimize the control and eliminate system errors presented by environment and internal noises to be able to implement autonomous navigation efficiently. 1.2.2. Particulate Matter 2.5 (PM2.5) In other studies, mobile robot is used to do the tasks mentioned above. In this paper, the main objective of the mobile robot is to navigate around an indoor environment to map the concentration of particulate matter 2.5 (PM2.5) and filter the air whenever there is a high concentration of the said particle. An air filtering device is integrated with the mobile robot to enable the air filtering part.
This air filtering device is automatically controlled by the PM2.5 sensor so that only whenever there is a high concentricity of particulate matter that it will be turned on in order to save battery life for longevity of operation time. Additionally, the concentration of the PM2.5 on the given environment is mapped with the help of the StarGazer indoor positioning system. The map is created on real-time basis on the remote computer. The study aims to improve the healthcare and safety standards inside a work or indoor environment by monitoring the air quality and at the same time negate the hazardous effects of the PM2.5. This will also minimize the work effort to monitor the air quality daily by having an autonomous mobile robot doing the monitoring itself on its own. In recent years, several studies have been conducted to also monitor the PM2.5 concentricity.
In 27, Piyavach Khunsongkiet and Ekkarat Boonchieng proposed a mobile interface to convert air quality monitoring by using a low cost sensor data to a digital value. This study was conducted because particulate matter pollution was becoming a catastrophic problem in many countries around the world. People spending a lot of time outdoor have to face the effects of particle pollution in the atmosphere. Therefore, they design a device that would help these people to monitor where the concentricity of the PM2.5 ¬is very high.
This would mean that they would have to go to the said environment first to verify the data. Another study about PM2.5 focused on developing smart cities where population was starting to increase rapidly and air pollution was becoming a major issue from public health to social economy. To enhance the quality of urban living, a participatory urban sensing framework for PM2.5 was presented with more than 2500 devices deployed in Taiwan and 29 other countries 28. All these research efforts was conducted to improve the overall quality of healthcare system of a certain community or country. In this paper, we optimize the use of a mobile robot by doing the task of autonomously mapping the status or concentration of PM2.5 ¬and at the same time filtering the air ¬in an indoor environment. In this case, a human operator will not be needed anymore; therefore, avoiding the hazardous effects of the particulate matter to humans.
This study would be beneficial to certain environments where the particulate matter’s concentration is very high that it would be very dangerous for humans to go or work in to. 1.3. Organization of Dissertation Chapter 1 – Introduction This chapter shows the research motivation of the paper, some literature review, and briefly introduces the overall system structure. Chapter 2 – Introduction to Mobile Robot This chapter introduces the used mobile robot. The kinematics of two-wheeled mobile robots will be also discussed in this chapter. Chapter 3 – Introduction to PM2.5 Sensor The PM2.5 sensor that was used is presented in this chapter. The microcontroller that was used to operate the sensor will also be discussed in this chapter.
Chapter 4 – System Architecture, Control Principles and Methodology The overall system architecture is discussed in this chapter. The data acquisition, data planning and data processing is also further discussed. Chapter 5 – Experiment Results Experiment results of the several controllers used in doing the wall following and obstacle avoidance is shown in this chapter. The trajectory of the mobile robot and the step response time is discussed to prove the effectiveness of the proposed dual-mode controller. The map of the concentricity of the particulate matter that is produced using the StarGazer and the PM2.5 sensor is also shown in this chapter.
Chapter 6 – Conclusion The final analysis and some limitations is discussed. Chapter 2 Autonomous Mobile Robot Pioneer 3-DX is a sturdy but lightweight two-wheeled mobile robot that has eight sonar transducers in front that provides object detection and range information. The sonar positions in the mobile robot is fixed: one on each side, and six facing outward with 20-degree intervals. Together, fore and aft sonar transducers provide 360 degrees of nearly seamless sensing for the platform. The sonar transducers firing pattern is from 0 to 7 or from left to right as shown in Fig.
2-1. Figure 2-1. Pioneer 3-DX mobile robot’s physical dimension and sonar transducers’ orientation n m Figure 2-2. Pioneer 3-DX Mobile Robot 2.1.
Kinematics of a Mobile Robot A kinematic model of the considered mobile robot is presented on Figure 2-3. The robot configuration is represented by its position on the Cartesian space (x and y, that is the position of the mobile robot-body center with relation to a referential frame fixed on the workspace), and by its orientation ? (angle between the mobile robot orientation vector and the reference axis – X, fixed on the workspace). t Figure 2-3. A model of a two-wheeled mobile non-holonomic robot.
The kinematic model represents the movements constraints of the robot body. For the considered mobile robot, the kinematic model is given by (1). The vector q = represents the linear and angular positions and the vector q = represents thxe linear and angular velocities. The main features of this model for wheeled mobile robots is the presence of non-holonomic constraints, due to the rolling without slipping condition between the wheels and the ground. The non-holonomic constraints impose that the system generalized velocities cannot assume independent values.
It can be observed that in (11), dynamic effects of the mobile robot’s body and actuators are not considered. 2.2. Sonar Sensors Sonar (originally an acronym for SOund Navigation And Ranging) is a technique that uses sound propagation (usually underwater, as in submarine navigation) to navigate, communicate with or detect objects on or under the surface of the water, such as other vessels. A sonar sensor is a device that can measure the distance from an object by the use of sound waves. It can measure distance by sending out a sound wave at a specific frequency and listening for that sound wave to bounce back.
By recording the elapsed time between the sound wave being transmitted and received, it is possible to calculate the distance between the sonar sensor and the object. In Figure 2-4, the principle of an active sonar is shown. We ca use (2) to solve the distance between the object and the sonar sensor by using the speed of sound constant. In this paper, the data from the sonar sensors is acquired and displayed using MATLAB via wireless bluetooth interface with the mobile robot’s on board microcontroller. The sonar’s firing pattern is from left to right as stated above. The firing frequency can be adjusted according to the operator’s preference or mobile robot’s navigation type.
The data displayed on MATLAB is shown in Figure 2-5. Table 1 shows the specification of the used mobile robot. (2) Where d is the distance from the sonar sensor to the object. is the speed of sound which is 344 m/s (1129 ft/s) and t is the time taken for the sound wave to bounce back.
Figure 2-4. Principle of an active sonar Table 2-1. Pioneer 3-DX Mobile Robot Specifications Length 44cm Width 39cm Height 22cm Weight 9kg Payload 25kg Run Time 24hrs~30hrs Translate Max Speed 1400mm/s Wheel Diam. 195.3mm Wheel Width 47.4mm Swing 26.7cm Built-In Applications ARIA(C++API), JAVA API, MobileEyes Figure 2-5. Acquired data from sonar sensors Initially, the sonar frequency is set to 25 Hz; therefore, the sonar sensor data is acquired for one transducer for every 40 milliseconds. The first column from data the array acquired from MATLAB is in terms of millimeters. The second column is the number of sonar sensor of which gives data at the moment.
The sampling rate of the soanr sensor can be adjusted through MATLAB. Figure 2-6. Firing rate of sonar sensor The data acquired from 8 sonar sensors S0 S1 S2 S3 S4 S5 S6 S7 is used strategically to implement the automatic navigation of the mobile robot to control the wheel velocities and the heading angle. In this paper, the S0 and S7 data is significant in acquiring the system error when implementing the wall following control. The system error of the two sonar sensors is defined as: (2) (3) Where d0 and d2 are set point values for the system error, which can be adjusted according to the to the user’s preference. The system error (2) and (3) becomes inputs for the PID controller, where the controller drives either on of the two errors to zero.
Other sensor values are also used to minimize the system error whereas S1, S2, and S3 are used for left wall following and S4, S5, and S6 are used for right wall following. The system error for left wall following is defined as the following: (4) (5) Where d3 and d4 are set point values for e3 and e4 respectively. The values of (4) and (5) become the antecedent part of the interval type-2 fuzzy logic controller for the left following part. The design of the controller is explained further in Chapter 4.. 2.2.1 Sonar Sensor Data Inaccuracies It is important to note that some objects or environments might not be detected by sonar sensors or they might give inaccurate data at some times. The reason behind this is because some objects are shaped or positioned in such a way that the sound wave bounces off the object, but are deflected away from the sonar sensor.
It is also possible for the object to be too small to reflect enough of the sound wave back to the sensor to be detected. Other objects can absorb the sound wave all together, like for example, cloth can absorb the sound wave which means that there is no way for the sensor to detect them accurately. These are important factors to consider when designing and programming a robot using a sonar sensor. One good example of many phenomenon that affects the sonar sensor data is the specular reflection effect. The specular reflection effect is a phenomenon whereas the sound wave from the transceiver does not bounce back and is reflected away from the sensor, giving a data that is not accurate with the real distance from the sonar sensor and the object.
The specular reflection effect scenario is given on Figure 2-7. Figure 2-7. Specular Reflection Effect The main reason of the sound wave being redirected away from the receiver is because it is not perpendicular to the sonar sensor orientation. If the incident angle ?, is greater than the half aperture of the sonar beam, ?/2, with respect to the sensor orientation, the sensor will fail to detect the returned beam because the sound wave is reflected away from the receiver. In this paper, the data inaccuracies is dealt by using a better fuzzy logic controller which can deal with certain uncertainties. The type-2 fuzzy controller has the element of footprint of uncertainty making them ideal for modelling certain types of uncertainty/noise which cannot be modeled by type-1 fuzzy controllers.
Chapter 3 Introduction to Indoor Positioning System and Particulate Matter 2.5 (PM2.5) Sensor 2.1. StarGazer Indoor Positioning System IPS is a system that determines an object or device’s current location through various kinds of signals. In this paper, the IPS that was used is Hagisonic’s StarGazer IPS, it analyzes reflected images from emitted infrared light that are beamed off from the passive landmarks that are attached to the ceiling. Each landmark is encoded with a fixed pattern to determine distance and orientation pattern and with a unique ID for localization. It provides high accuracy in position and heading angle, so it is ideal to use in determining a robot’s current position when navigating a certain environment. The StarGazer is composed of both an Infrared projector and an image processing unit.
It is also simple to operate because there is no need for synchronization or communication between the robot and the landmarks. There is also no need for battery or power supply for landmarks. Noises from the environment from environments have minimum effect on the output data. Figure 3 shows the StarGazer device. Figure 4 shows the landmark that is attached to the ceiling.
Figure 3-1. Hagisonic StarGazer Indoor Positioning Device Table 3-1. Hagisonic StarGazer Specifications Hardware Interface UART(TTL 3.3V), 115,200bps Size 50mm × 50mm × 28mm Communication Protocol User protocol based on ASCII code Measurement Time 10 times/sec Localization Range (per a Landmark) 2.5~3m in diameter (for ceiling height 2.4m) Repetitive Precision 2 cm Heading Angle Resolution 1.0 degree Landmark Types (Classification for height range) HLD1: 1.1 ? height ? 2.9 m HLD2: 2.9 ? height ? 4.5 m HLD3: 4.5 ? height ? 6.0 m Landmark Types (Classification for total ID numbers) HLDnS: 31 ea (for a normal space) HLDnL: 4,095 ea (for a larger space) (n=1,2,3; see the classification for height range) Power Consumption 5 V: 300 mA, 12 V: 70 mA Table 3. Landmark Types and Height Range Model Height Range HLD1 1.1m?Height?2.9m HLD2 2.9m?Height?4.5m HLD3 4.5m?Height?6.5m Figure 3-2. Passive Landmark attached to the ceiling. 3.1.1.
Operation of Indoor Positioning System The data from the StarGazer indoor positioning device can be acquired through MATLAB which is connected via Bluetooth with the StarGazer. To acquire the the data from the StarGazer, the connection between the remote computer and the device must be secured first. To secure the connection between the remote computer and the device, they must run in the same baud rate and computer port initially. This can be set up in the StarGazerMonitor application.
Figure 3-3. Actual setup of the StarGazer Indoor Positioning Device Figure 3-4. StarGazerMonitor UI. The computer serial setting is boxed for setting up the correct computer settings.
3.1.2. Map Building Mode Figure 3-5. The data is acquired by the StarGazerMonitor If the Computer Serial Setting is set correctly, the StargazerMonitor can already acquire the data from the Stargazer through the CalcData. In this figure’s CalcData, 310 represents the reference ID Landmark that the Stargazer detects. • -69.15 indicates the value of angle (-180 to 180) • +18.49 is the position in the X axis with respect to ref.
ID in cm • -194.84 is the position in the Y axis with respect to ref. ID in cm • 204.37 is the position on the Z axis. (Height in cm) Figure 3-6. Map building mode – the landmarks were already detected and displayed in the StarGazerMonitor Note that the connection of the Stargazer and computer in the map building process can only be by wired serial communication and cannot be by Bluetooth. After the Stargazer gets the 6 landmark which is set in the StargazerMonitor ‘Num of Landmark’, the Stargazer automatically saves the map. 3.1.3.
Data Acquisition Figure 3-7. Map and data acquired from MATLAB Figure 3-8. Experimental scenario – minimum light In this simulation, the environment has minimum light. So the data that we get is very good. However, when there is so much light near the landmarks, the data we get has some pattern misrecognition.
Figure 3-9. Experimental scenario – light is abundant Figure 3-10. Data acquired when light is abundant 3.2. Particulate Matter 2.5 (PM2.5) Sensor The dust sensor used in this paper is Sharp GP2Y0. This reliable sensor measures the concentration of PM2.5 (PM2.5 refers to particles that are 2.5 microns or smaller in diameter) in a certain environment or room. This sensor uses laser scattering to radiate suspending particles in the air, then collects scattering light to obtain the curve of scattering light change with time.
The microprocessor calculates equivalent particle diameter and the number of particles with different diameter per unit volume. Figure 3-11 shows the dust sensor that is used in this paper. Figure 3-13 shows the pin configuration for each wire. Figure 3-11. Sharp GP2Y10 Dust Sensor Figure 3-12.
Sharp GPY10 Dust Sensor Pin Assignments Figure 3-13. Sharp GPY10 Dust Sensor Pin Configuration Table 3-2. PM2.5 Sensor Pin Definition Pin Function V-LED Connect to 5.0 Vwith resistor of 150 ohms in between. LED-GND Connect to GND. LED Connect to any digital pin of Arduino board.
S-GND Connect to GND. Vo Connect to any analog pin of Arduino board. Vcc Connect to 5.0 V. 3.2.1. Microcontroller (Arduino Uno) Using the Arduino Uno microcontroller, we can extract the data from the Sharp GP2Y10 PM2.5 Sensor. Also, this microcontroller is used to control the operation of the air filter when there is a high concentricity of PM2.5.
The connection of the dust sensor and air filter to Arduino Uno is shown in Figure 15. The whole is system is powered by a 5 volts battery. Figure 3-14. Experiment no. 1 – no object is inserted in dust detection area Figure 3-15. Experiment no.
2 – an object is inserted in the dust detection area Figure 3-16. Initial data from the dust sensor using Arduino Uno from experiment 1 Figure 3-17. Initial data from the dust sensor using Arduino Uno from experiment 2 Looking at Figures 3-16 and 3-17, we can see that the data acquired is categorized into three. Namely, Raw Signal Value, Voltage Measured, and Dust Density.
The dust density, which is measured by µg/m³ is acquired from the two data categories. We can observe that by using the Arduino Uno microcontroller, when simulating a completely polluted air by inserting an object in the dust detection area, we max out at 411.00 value for the Raw Signal Value, and 1.32 for the Voltage Measure, and 0.13 µg/m³ for the dust density. Having only reached 411.00 acquired for the Raw Signal Value, we consider the results of the experiments unacceptable. Therefore, further experiments was done to meet the acceptable values for the dust density. Chapter 4 Methodology 4.1. Interval Type-2 Fuzzy-PID Dual-Mode Controller For the AGV’s operation control, the interval type-2 FLC which has the advantages of meeting nonlinear and time-variant controller objects is used for the large curvature path tracking and obstacle avoidance.
In contrast, the traditional PID controller which has higher control precision and eliminates system error for tracking straight paths and small curvatures effectively. The idea of the IT2FPIDDMC is to have a changeover switch to implement a controller according to the current situation of the mobile robot or according to the current situation of the environment. The system error which can be obtained through the sonar sensors is processed and used as a metric to control the mobile robot’s right and left wheel velocities. The switching threshold K is a constant value which is initially set to 450mm in this paper according to the experiments. The switching threshold K can also be adjusted according to the user’s preference.
Figure 4-1. The schematic diagram of the IT2FPIDDMC 4.2. Interval Type-2 Fuzzy Logic Controller The fuzzy logic controller is illustrated in Figure 8. The controller has a pre-established fuzzy controller, which consists of four components: fuzzification, fuzzy inference rule base, and defuzzification. IT2FLC works as follows: when the switching threshold K is larger than S1, S2, and S3, (4) and (5) is used as inputs for the fuzzy logic controller, whereas they are converted from real numbers to fuzzy values.
The fuzzy inference engine then processes the input values based on the pre-defined rule base to compute the corresponding control outputs. In this paper, the fuzzification process for (4) and (5) is defined respectively by triangular type-2 MFs which is shown in the following: Figure 4-2. Membership function for (4) Figure 4-3. Membership function for (5) Figure 4-4. Key components of FLC system In order to implement both right and left wall following, the robot must optimize all the sonar sensors; In the previous chapter, only the left wall following is explained and S4, S5, and S6 is not used.
The system error for these three sensors can be described according to the following: (6) (7) Where d5 and d6 are set point values for e5 and e6 respectively. The values of (6) and (7) become the antecedent part of the interval type-2 fuzzy logic controller for the right following part. The membership function of e5 and e6 is shown in the figures below: Figure 4-5. Membership function for (6) Figure 4-6.
Membership function for (7) The fuzzy inference rule base of the IT2FLC is pre-defined and is carefully designed through series of experiments. The design of the rule base is shown in Table 3. Table 4-1. Fuzzy inference rule base of running velocity Rule e3 e4 VL VR 1 NE NE F B 2 NE ZE F Z 3 NE PE F Z 4 ZE NE F B 5 ZE ZE F F 6 ZE PE Z F 7 PE NE F Z 8 PE ZE Z F 9 PE PE Z F 4.3. Type-2 Fuzzy Logic System A type-2 fuzzy set à is characterized by a type-2 membership function µÃ (x, u) 29, where x ? X and µ ? Jx ? 0, 1, i.e., à = {((x, u), µÃ (x, u)) | ? x ? X ? u ? Jx ? 0, 1} (8) in which 0 ? µÃ (x, u) ? 1. à can also be expressed follows 29: à = JX ? 0, 1 (9) where ? ? denotes union over all admissible x and u 29.
Jx is the primary membership of x, where Jx ? 0, 1 for ? x ? X 29. The uncertainty in the primary memberships of a type-2 fuzzy set Ã, consists of a bounded region that is called footprint of uncertainty (FOU). It is the union of all primary memberships 29. It has been proven that regardless of the primary membership function, the resulting FOU is about the same 29. The ith rule of the type-2 fuzzy logic system can be interpreted in the following form: Rule : IF x1 ¬is AND.
. . AND xn is THEN y is , = 1, . .
. , M (10) Where , j = 1, . . . n, is an interval type-2 fuzzy set, ? ? is a crisp value , and M is the number of rules. The consequent control action in this paper is chosen from a predefined control action set U, which is based on the rule table designed in this paper.
There are many types of membership function that can be used in a fuzzy logic controller. For type-1 fuzzy logic system, trigonometric membership functions can be represented by: (11) Where ?(x) is the membership degree, a is the lower limit, m is the modal value, and b is the upper limit. In type-2 fuzzy logic system, the FOU of this MF can be represented as a bounded interval in terms of upper MF (x) and lower MF (x), where: (12) (13) Where N is a numerical constant to determine the membership degree of the lower membership function. The computation of the fuzzy controller output y consists of fuzzification, inference engine, type reduction, and defuzzification 30.
In the fuzzification, crisp input x = (x1, . . . xn) is converted into a fuzzy singleton and is mapped to a fuzzy set with an interval degree of the (5) and (6). The inference engine used in this paper is implemented by fuzzy t-norm operation using the algebraic product. The rule firing strength , which is an interval type-1 fuzzy set is computed as follows 31: (14) where (15) While many type-reduction processes have been proposed, like the one, which is most widely used, the center of-sets type-reduction method, this paper uses a simplified type-reduction, which is called upper and lower extreme type-reduction.
Instead of considering all the upper and lower firing strengths of rules, this process uses only two specific combinations which is described as follows 32: (16) The final output of type-reduction is a type-1 fuzzy set comprising of two fuzzy singletons at and 33. This simple type-reduction process avoids a complicated iterative computing and is much simpler to implement. The final defuzzification process finds the centroid of (31), which is simply the average of the two fuzzy singletons. We get the final output in the form of 32 : (17) 4.4.
Proportional Integral Derivative Controller (PID Controller) PID controller is a widely used three term controller which is basically a controller with a control loop feedback mechanism. This controller has been around the control system industry and various other applications requiring continuously modulated control for so many years because of its efficiency and design simplicity. A PID controller calculates an error value e(t) as the difference between a certain desired set point value and a measured process value and applies the correction on the system error based on proportional, integral, and derivative terms (denoted P, I, and D respectively). In this paper, the PID controller is used to track path with straight lines in doing the wall following and obstacle avoidance system. A PID controller which has a higher control precision than a fuzzy controller, is chosen as a part of the dual-mode controller because it can eliminate system error for tracking paths with straight lines with small curvatures easily.
The overall control function of the PID controller used in this paper can be expressed mathematically as: (17) Where is the initial control state or the feed forward term and can be expressed as: (18) Where adjusting the Kp, Ki, and Kd gains can reasonably implement effective control. For example, increasing Kp can improve response speed and reduce static error; however, it may lead to overshoot and oscillation. The system architecture of the PID controller used is shown in Figure 11. Figure 4-7.
PID controller system architecture In this case, the e(t) used in the system is (2) and (3). The feed forward term must be expressed in terms of velocity/second in order to implement a control output of turning angle or running velocity. Several experiments have been done to show the performance of the PID controller which is shown in Chapter 5. 4-5.
Type-1 Fuzzy Logic Controller A fuzzy controller has four components, namely, the fuzzification interface linguistic rule base, inference engine, and the defuzzification interface. In fuzzification, crisp inputs are tarnsformed into fuzzy inputs. In order to do so, the universe of discourse of the crisp input have to be assigned and membership functions are defined for each input. The knowledge of how to control the system best is embedded in the linguistic rule-base in the form of IF-THEN rules. The inference engine is the one that evaluates which control rules are more relevant and then determines the corresponding action that should be taken.
Figure 4-8. System architecture of the type-1 fuzzy logic system 4.5.1. Fuzzification Fuzzification is the process of assigning linguistic values to the inputs. The input space will have a several sets of elements and then these elements will be defined by its membership function. The fuzzy quantity will be defined by its membership degree.
Membership degree defines the level of each element that is attached to the fuzzy set, and it should be a numerical value that is between 0 and 1. There are many types of membership function that can be used in a fuzzy logic controller. In this paper, trigonometric and trapezoidal is exclusively used as Where ?(x) is the membership degree, a is the lower limit, m is the modal value, and b is the upper limit. Trapezoidal Function Where a and b is the two inflection points on the bottom of the trapezoid, while c and d are the two inflection points on the top. To acquire the input variables which will be fuzzified, the eight sonar sensors of Pioneer 3-DX was group into three, {S0 S1 S2, S3 S4, S5 S6 S7}.
Namely, left sonar sensors, front sonar sensors, and right sonar sensors. The minimum distance acquired from each group is then defined as the input variables. Using the data acquired from the sonar sensors, we can determine the mobile robot’s distance to the wall and define the fuzzy subset as {Near, Quite-Near, Far}. Setting the input space to 0 to 1000 millimeters, the membership distribution is shown in Figure 12. The output of the fuzzy controller is the speed of right and left wheel velocity, we set its input space first to -100 to 100 millimeter per second.
The input space of the output is changed in some experiments to determine the effectiveness of the fuzzy controller even if we vary the speed. The fuzzy subsets are {BF, B, Z, F, FF}, which means backward fast, backward, zero, forward, and fast forward respectively. Figure 4-9. Membership Function for Sonar012, Sonar34, Sonar567 4.5.2. Fuzzy Rules Determining the fuzzy rules for a fuzzy logic controller obtains from doing a lot of experiments, objectives or observation on expert human operators.
The number of rules is actually determined by the number of input variables as the rules increases exponentially when the input increases. In order to simplify the design and efficiency of the fuzzy logic controller, reducing the rules as much as possible is taken into consideration without affecting the objective and effectiveness of the controller. This paper describes the fuzzy logic controller to implement a wall following and obstacle avoidance to a mobile robot. So the objective of the controller is to find the nearest wall first and follow its outline. The output of FLC is the velocity of right and left wheel, so in order for the mobile robot to follow a trajectory along the outline of the wall and avoid obstacles, the rules for their linear velocity must be carefully taken into consideration that will allow the mobile robot to imitate the desired position vector. The fuzzy rules used in this paper is listed in Table 4.
Table 4-2. Fuzzy Controller Fuzzy Rules Rule Sonar012 Sonar34 Sonar567 LWV RWV 1 N N N BF BF 2 N N QN FF BF 3 N N F FF BF 4 N QN N F F 5 N QN QN FF Z 6 N QN F FF BF 7 N F N F F 8 N F QN F B 9 N F F FF BF 10 QN N N BF FF 11 QN N QN BF FF 12 QN N F FF BF 13 QN QN N B FF 14 QN QN QN Z FF 15 QN QN F FF B 16 QN F N B F 17 QN F QN Z FF 18 QN F F FF Z 19 F N N BF FF 20 F N QN BF FF 21 F N F BF FF 22 F QN N B FF 23 F QN QN Z FF 24 F QN F Z FF 25 F F N Z FF 26 F F QN F FF 27 F F F FF FF To explain some of the rules: Rule 27: If all the sonar sensor’s data is “far” which is defined in out input fuzzy membership function. The mobile robot’s objective is to continuously moving forward until it finds a wall to follow or an obstacle to avoid. Rule 19: If the sonar sensor group on the front and right detects a “near” obstacle which is also defined in our fuzzy membership function, the mobile robot’s objective is to turn large to the left to avoid the obstacle in front but maintain a distance that is not too far from the right wall. Expanding the rule table or increasing the rules by increasing the input, like using all the sonar sensors as inputs was considered in making the fuzzy logic controller, however, having eight inputs in a fuzzy logic controller will increase the rule exponentially and it might require longer computation time and might cause degradation in fuzzy logic controller’s performance.
Hence, other methods can be studied in the future to improve the FLC performance. 4.5.3. Defuzzification Defuzzification converts the fuzzy quantity into actual control values. There are several defuzzification methods that is available, but in this paper we used the center of gravity for the type-1 fuzzy controller which is considered to be the simplest and most useful defuzzification method. To show the basic operation on how to acquire the crisp data outputs by using the center of gravity defuzzification method.
Some example input data is shown in Figure 14 to acquire the corresponding crisp output using Table 3. In Figure 14, rules 1 and 5 only has rule strength to have a corresponding output. The inputs is weighted or aggregated using an AND operator to identify the weight of the output that will be used in the center of gravity method. Figure 4-10.
Center of gravity defuzzification method Another method for acquiring crisp outputs is center of sets defuzzification method using the Sugeno FIS. This method of defuzzification is much less complex than center of gravity method which requires more computation time which is shown in Figure 5-11. Figure 4-11. Center of sets defuzzification method Chapter 5 Automated Air Filtering and PM2.5 Real-Time Mapping System Automatic air filtering and real-time mapping system is realized by combining the data from the PM2.5 sensor and the data from the StarGazer indoor positioning system. The data is synchronized through MATLAB and the real-time mapping is also done in the same software.
The idea of the whole system is to operate the air filtering system only when there is high concentricity of PM2.5 to save the battery life of the mobile robot. Also, the dust density is mapped in real-time to show the areas which has high concentricity of the said PM2.5 so the people in the area will have an idea of the safe and dangerous locations in their environment. The goal of this study is to provide an autonomous mobile robot that can provide a clean air in a certain indoor environment and at the same time, map the PM2.5 concentricity in the same indoor environment in real-time. The mobile robot that is used is the P3DX mobile robot which has built-in 8 sonars sensors that gives the robot an idea of the surroundings and navigate by using the data that it acquires. The air filtering and indoor positioning system is integrated with the AGV system; therefore, we can know the current position of the mobile robot and know the concentricity of PM2.5 in that specific area at the same time as well.
This will then gives a an idea to operate the air filtering system in only specific locations that PM2.5 ¬concentricity is very high. To determine the concentricity of PM2.5 in certain regions, color distribution is used in the map. Whereas, green is the safe region where the PM2.5 concentricity is from 0.00-0.11 µg/m³. The blue region represents the quasi-safe region which the concentricity is from 0.12-0.22 µg/m³. Finally, the red region represents the dangerous region which the concentricity of PM2.5 is from 0.23 µg/m³ and above.
The automatic air-filtering and PM2.5 real-time mapping system is made possible by simultaneously acquiring the data from the PM2.5 sensor and the StarGazer indoor positioning system. The setup of the system is shown in Figure 5-1. The map acquired from the first and second run is shown in Figure 5-2 and Figure 5-3, respectively. Figure 5-1.
Mobile robot with air-filtering and PM2.5 mapping system Figure 5-2. Map acquired from the first run of the mobile robot Figure 5-3. Map acquired from the second run of the mobile robot It can be observed in the figures above that most of the regions in the experimental environment can be defined as safe regions, while on the other hand, some region shows that it is in quasi-safe and dangerous region. Therefore, it can be concluded that in an indoor environment, PM2.5 ¬concentricity also varies in small scales and it can be mapped in order to give information of how safe a certain location in a certain indoor environment is. Also, the mobile robot’s filtering system will be operating only on the quasi-safe and dangerous region. In regions where the feedback data from the PM2.5 sensor is below 0.11 µg/m³, the air-filtering system will be turned off.
Also, to determine the exact concentricity of the PM¬2.5 in each exact current coordinate of the mobile robot, the data is acquired in MATLAB. The data acquired is shown in Figure 5-4. Figure 5-4. PM2.5 concentration from each coordinates data In the figure above, the coordinate data Xand Y is in the leftmost side which is in the bracket. The numerical data in the rightmost side shows the concentration of PM2.5, which is averaging on 0.25 µg/m³. In several experiments, maximum data acquired with most extreme conditions is 0.32 µg/m³ and the lowest data acquired is 0.05 µg/m³.
Some acquired data is shown in Table 5-1. Table 5-1. Data acquired from the system X – coordinate (k) (mm) Y – coordinate (k) (mm) Dust density (k) (µg/m^3) 0.00 0.00 0.05 91.00 34.00 0.05 152.00 106.00 0.05 158.00 98.00 0.06 -2889.00 -1733.00 0.06 -3162.00 -1859.00 0.06 156.00 101.00 0.08 75.00 175.00 0.08 15.00 27.00 0.07 166.00 140.00 0.05 -92.97 -263.08 0.24 -35.74 -242.95 0.25 2.72 -428.83 0.25 53.23 -493.71 0.32 -42.77 -550.11 0.32 Chapter 6 Experiment and Results In this chapter, all the controllers that was presented on the previous chapters were used and experiment results are shown accordingly. The efficiency of the IT2FPIDDMC over the other controllers is proven through step response curves and trajectory mapping. Also, the map of the concentration of the PM2.5 created through the indoor localization system and the PM2.5 sensor is presented in this chapter. Experiment results shows that the concentration of the particulate matter can be mapped effectively and autonomously by the autonomous guided vehicle.
The following figures shows the experimental scenarios when the simulation is conducted. Note that the following step response curves that is presented in this chapter is acquired through (2) and (3) Figure 6-1. Experiment scenario 1 – mobile robot following the inside contours of a structured wall Figure 6-2. Experimental scenario 2 – mobile robot following the inside contours of an unstructured wall Figure 6-3.
Experimental scenario 3 – mobile robot following the outside contours of a structured wall Figure 6-4. Experiment scenario 4 – mobile robot following the outside contours of an unstructured wall 6.1 Type-1 Fuzzy Controller Simulation Results Using a type-1 fuzzy controller for the wall following and obstacle avoidance system for the P3-DX mobile robot to navigate a certain environment autonomously, experiment results show that it was able to deal with sharp curves or non-linear systems, although having difficulties in tracking paths with straight lines. Figure 6-5. Experimental scenario 1 type-1 fuzzy logic controller trajectory Figure 6-6. Experimental scenario 2 type-1 fuzzy logic controller trajectory Figure 6-7.
Experimental scenario 1 step response curve using type-1 fuzzy controller Figure 6-8. Experimental scenario 2 step response curve using type-1 fuzzy controller Figure 6-9. Experimental scenario 3 step response curve using type-1 fuzzy controller Figure 6-10. Experimental scenario 4 step response curve using type-1 fuzzy controller We can observe in Figure 6-8, where the experimental scenario is following the outside contours of a structured wall where there are no obstacles present by using a type-1 fuzzy controller that the error value is quiet unstable. The error value in this experiment reached up to 2.911 using equation (2).
Although in Figure 6-9, where the experimental scenario is the same, whereas the mobile robot would have to follow the outside contours of the wall but with obstacles present; we can observe that the maximum error value that was reached was 5.753. Hence, proving that unstructured environments would have some effect of the controller’s robustness. Figures 5-10 and 6-11 shows experimental scenario 3 and 4, respectively; the mobile robot would have to follow the inside contours of a wall with experimental scenario 4 having obstacles in the environment. To further explain the type-1 fuzzy controller’s performance, the left and right wheel velocity’s step response curve is shown in Figure 6-12. It can be observed that velocity response is quiet unstable when using the type-1 fuzzy controller.
Further improvements in the velocity response is shown when PID controller is added in the system. Figure 6-11. Left and right wheel running velocity step response curve of type-1 fuzzy controller 6.2. PID Controller Simulation Results Alternatively, PID controller was also used to test its effectivity in implementing wall following and obstacle avoidance system.
It can be seen on the experiment results that it has a high efficiency in dealing with tracking path with straight lines but was having difficulty in dealing with non-linear system and non-structured environments. Figure 6-12. Experimental scenario 3 PID controller trajectory Figure 6-13. Experimental scenario 1 step response curve using type-1 fuzzy PID controller Figure 6-14. Experimental scenario 2 step response curve using type-1 fuzzy PID controller Figure 6-15.
Experimental scenario 3 step response curve using type-1 fuzzy PID controller Figure 6-16. Experimental scenario 4 step response curve using type-1 fuzzy PID controller By introducing the PID controller and integrating it in our controller, we can observe that the repose of the system error is much better that the fuzzy controller’s response alone. As shown in Figure 6-14, we can observe that the response of the controller to reach a 0 value system error is faster even when our initial system error is quite large. From this, we can also see that the system error maintained a value that is quite close to 0. We can also observe in Figures 6-15 and 6-17 whereas the experimental scenarios are simulated in unstructured environments that even though the step response curve is unstable unlike in experimental scenarios 1 and 3, the response of the system error is still better than the type-1 fuzzy controller’s performance; hence, we can conclude that the type-1 fuzzy PID controller’s performance is better than the type-1 fuzzy controller’s performance.
Figure 6-17. Left and right wheel velocity step response for Type-1 Fuzzy-PID controller 5.3 IT2FPIDDMC Simulation Results The objective of the design of the IT2FPIDDMC is to optimize the autonomous navigation capability of the mobile robot. The IT2FPIDDMMC has a better step response curve than the two previous controllers by optimizing the best characteristics of the said two cotnrollers. It can be seen in the experiment results that the IT2FPIDDMC can deal with both non-linear and linear systems effectively. Figure 6-17 and 6-18 shows the experimental scenarios. Figure 6-18.
Interval type-2 fuzzy PID dual-mode controller trajectory Figure 6-19. Experimental scenario 1 step response curve using interval type-2 fuzzy PID dual-mode controller Figure 6-20. Experimental scenario 2 step response curve using interval type-2 fuzzy PID dual-mode controller Figure 6-21. Experimental scenario 3 step response curve using interval type-2 fuzzy PID dual-mode controller Figure 6-22. Experimental scenario 4 step response curve using interval type-2 fuzzy PID dual-mode controller The step response cruve of the Interval Type-2 Fuzzy-PID Dual-Mode Controller can be seen in Figures 6-20 to 6-23.
It can be observed that the step response of tge IT2FPIDDMC is better that the previous two controllers. It can be observed that is has faster response and better stability than the other two. Although the response curve acquired from experiments with unstructured environments still have some high value errors acquired that goes up to 0.510, the performance is still better that the type-1 fuzzy-PID proving that the type-2 fuzzy controller adds robustness to the system. Additionally, as shown in Figure 6-24, the velocity response of the mobile robot is much better than the two previous controllers. Figure 5-23. Left and right wheel velocity step response for interval type-2 fuzzy PID dual-mode controller Chapter 7 Conclusion and Future Works In this paper, interval type-2 fuzzy logic controller was used to implement autonomous navigation for mobile robots.
It was integrated with a conventional PID controller to form a dual-mode controller which can be switched over by a certain switching threshold. The interval type-2 fuzzy controller was developed in order to deal with uncertainties that is presented by unstructured and changing environments. Also, internal noises that cause the sensor data inaccuracies can be dealt with the FOU of the type-2 fuzzy membership functions. Experiment results show that the IT2FPIDDMC was more capable of navigating around in an unknown environment effectively than previously developed controllers. The mobile robot was able to follow the wall with good stability like a PID controller and was able to avoid obstacles like the fuzzy controller. Reference 1 Itai Kloog “Fine particulate matter (PM2.5) association with peripheral artery disease admissions in the northeastern United States” in Epidemiology, Vol.
26, pp. 572-277, June 2016 2 Chen, Hsiang-Ching, Ku, Chih-Hung “characteristics of pm2.5 in Taiwan’s largest cities during 2006–2009” in epidemiology, vol. 22, January 2011 3 Wang Min Ling “Breathing the Air We don’t Know: The Problem of PM2.5” 4 Yee-Lin Wu, Jhih-Siang Jian, Jhong-You Kel, and Chen-Chieh Kuo “Identification of the Impact of Dust Storm on the Ambient PM10 Concentrations in Southern Taiwan” 5 Sungchul Kang, Woosub Lee, Kyungchul Shin, Munsang Kim “ROBHAZ-rescue: rough-terrain negotiable tele operated mobile robot for rescue mission” in International Workshop on Safety, Security and Rescue Robots, pp.105-110, June 2005 6 W.K. Fung, Y.Y. Leung, M.K.
Chow, Y.H. Liu, Y. Xu, W. Chan, T.W.
Law, S.K. Tso, C.Y. Wang “Development of a hospital service robot for transporting task” in International Conference on Robotics, Intelligent Systems and Signal Processing, pp.628-633, October 2003 7 Z. Riaz, A. Pervez, M.
Ahmer, J. Iqbal “A fully autonomous indoor mobile robot using SLAM” in International Conference on Information and Emerging Technologies (ICIET), pp. 313-317, June 2010 8 Adarsh S, Mohamed Kaleemuddin S, Dines Bose, K I Ramachandran “Performance Comparison of Infrared and Ultrasonic Sensors for Obstacle of Different Materials in Vehicle/Robot Navigation Applications” in IOP Conf. Series: Materials Science and Engineering, Vol. 1, pp. 1-8, 2016 9 R.
Alba-Flores, F. Rios-Gutierrez, C. Jeanniton “Qualitative evaluation of a PID controller for autonomous mobile robot navigation implemented in an FPGA card” in International Conference on Natural Computation (ICNC), Vol. 7, pp. 233-238, September 2011 10 Abdullah Al Mamun Khan, M. Sultan Mahmud Rana, Jubayer Alam Rabin, Abu Farzan Mitul, Shahjahan “Design and Implementation of a Robot for Maze-Solving with Turning Indicators Using PID Controller” in International Conference on Informatics, Electronics, and Vision (ICIEV), pp.
211-217, August 2013 11 Cheng-Wen Lee “Green Suppliers Assessment Using Fuzzy AHP” in Journal in Advance Engineering, Vol. 4, No. 2, pp. 192-202/April 2009 12 Min-Chan Hwang “The Fuzzy Sets with Continuous Membership Functions in Topological Spaces” in Journal in Advance Engineering, Vol.
4, No. 2, pp. 147-150/April 2009 13 Yean-Ren Hwang, Shih-Yao Huang “Design FPGA Controllers for Air Motor via MRAC and Fuzzy Theory” in Journal in Advance Engineering, Vol. 4, No. 1, pp.
63-69/January 2009 14 Ying-Jih Chao Lee “Appraisal Performance of FDI Strategy in China – AHP, Fuzzy Theory and TOPSIS Methods” in Journal in Advance Engineering, Vol. 9, No. 1, pp. 1-8/January 2014 15 Siripun Thongchai, Kazukio Kawamura “Application of Fuzzy Control to a Sonar-Based Obstacle Avoidance Mobile Robot” in Proceedings of IEEE International Conference on Control Applications, pp. 425-430, 2000. 16 Hasan A.Yousef “Fuzzy-Logic Obstacle Avoidance Control: Software Simulation And Hardware Implementation” in European Control Conference, pp.
1458-1463, 1999. 17 H. Mamdani, S. Assilian, “An Experiment in Linguistic Synthesis With a Fuzzy Logic Controller”, in Int. Journal of Man-Machine Studies, Vol.
7, no. 1, p. 1-13, 1975. 18 T. Takagi and M. Sugeno, “Derivation of fuzzy control rules from human operator’s control actions”, Proceedings of the IFAC Symp. on Fuzzy Information, Knowledge Representation and Decision Analysis, p. 55-60, July 1983 19 L. A. Zadeh, “Outline of a new approach to the analysis of complex systems and decision processes”, IEEE Trans. on Systems, Man and Cybernetics, 3 (1): p. 28-44, Jan. 1973 20 J. Borenstein and Y. Koren, “Obstacle avoidance with ultrasonic sensors,” in IEEE Journal o,f Robotics, and Automation, Vol. 4, no. 2, April 1988. 21 R. E. King, Computational Intelligence in Control Engineering, Marcel Dekker, Inc., USA, 1999. 22 L. A. Zadeh, ”Fuzzy sets,” Information and Control, Vol. 8, pp. 338-353, 1965. 23 J. Mendel, Uncertain Rule-Based Fuzzy Logic Systems: Introduction and New Directions. Upper Saddle River, NJ: Prentice-Hall, 2001 27 P. Khunsongkiet, E. Boonchieng “Converting air quality monitoring low cost sensor data to digital value via mobile interface” in International Conference on Biomedical Engineering, pp. 57-62, February 2017 28 Ling-Jyh Chen, Yao-Hua Ho, Hu-Cheng Lee, Hsuan-Cho Wu, Hao-Min Liu, Hsin-Hung Hsieh, Yu-Te Huang, Shih-Chun Candice Lung “An Open Framework for Participatory PM2.5 Monitoring in Smart Cities” in IEEE Access, Vol. 5, pp. 131-137. July 2017 29 J. Mendel and R. John “Type-2 fuzzy sets made simple,” IEEE Trans. Fuzzy Syst., vol. 10, pp. 117-127, Apr. 2002. 30 J. Mendel, Uncertain Rule-Based Fuzzy Logic Systems: Introduction and New Directions. Upper Saddle River, NJ: Prentice-Hall, 2001 31 Q. Liang and J.M. Mendel, “Interval type-2 fuzzy logic systems: Theory and design,” IEEE Trans. Fuzzy Syst., vol. 8, no. 5, pp. 535-550, Oct. 2000. 32 C. F. Juang, C. H. Hsu “Reinforcement Ant Optimized Fuzzy Controller for Mobile-Robot Wall-Following Control,” in IEEE Transactions on Industrial Electronics, Vol. 56, no. 56, pp. 3931-3940, October 2009 33 Hani A. Hagras, “A Hierarchal Type-2 Fuzzy Logic Control Architecture for Autonomous Mobile Robots,” IEEE Transactions on Fuzzy Systems, vol. 12, no. 4, pp. 524-539, August 2004.
