(TITLE OF THE THESIS)* - UTPedia

INTRODUCTION

Introduction

In the last three decades, mobile robots have become a topic of considerable interest due to a wide range of possible applications. The research devoted to mobile robots started in three different domains; the first was the need for more flexible automated guided vehicles in manufacturing facilities, the second domain to stimulate research in mobile robots was planetary rovers and the third domain was the research in 'artificial intelligence', where mobile robots proved to be an attractive test bed for various artificial intelligence techniques [3]. Land-based wheeled mobile robots (WMRs) are increasingly present in industrial and service robots and are considered the most popular among researchers.

Note that the degree of mobility represents the number of degrees of freedom that can be used immediately, without reorientation of the orientable wheels.

Motivation of the Research

The planned path is usually decomposed into line segments between ordered subgoals or waypoints. Global path planning takes into account all the information in the environment when the optimum path ends between the start and target location. This thesis mainly studies the control and path planning issues of WMR in dynamic environments.

Attractive potentials are given to cells that are close to the goal of the mobile robot, while repulsive potentials are assigned to obstacles.

Research Objectives

Although the methods are fast, they can get stuck in the local minimum of the potential function [6]. To avoid local minima and at the same time to obtain the best optimal path, an automatic planning approach is needed.

Contribution of Thesis

Organization of the Thesis

LITERATURE REVIEW

Overview of Autonomous Mobile Robots

The robot's environment is usually designed according to the robot's task and then protected against external influences. Mobile robots are generally robots that can move from place to place on the ground. By their nature, non-holonomic mobile robots have fewer degrees of freedom than holonomic mobile robots.

These few actuated degrees of freedom in non-holonomic mobile robots are often independently controllable or mechanically decoupled, further simplifying the low-level control of the robot.

Table 2.1: Milestones in the evolution of robotic systems [5]

Functionalities of a Mobile Robot

Drive System of Wheel Mobile Robots (WMR)
Controller of Mobile Robots
Path Planning of Mobile Robot

A summary of the research conducted so far on controllers is presented in Table 2.3. The discussion in this thesis focuses on the global route planning problem of planning issues. The summary of research conducted so far on mobile robot path planning is presented in Table 2.4.

The performance of conventional GA (one criterion; minimum path length) was compared with that of MOGA (two criteria for a holonomic for a holonomic on a 2-dimensional network for minimum path length and difficulty). The simulation result found that conventional GA and MOGA show that both types of GA are effective tools for solving the point-to-point route planning problem.

Discussion

Based on the literature review and discussed earlier, it was found that differential operation with the Fuzzy Logic approach has scope for further investigation. It does not require heavy computation as some neural networks, and they are able to handle inaccuracy and uncertainties that are often present in many real-world problems. In addition, Ant Colony Optimization is easy to use and has the ability to search for optimal or near-optimal solution in the search area.

Therefore, ant colony optimization was adopted as an optimization technique for path planning in this research because it can provide explicit information and with less computation time.

Summary

MODELING AND SIMULATION

Modeling of Mobile Robot

Kinematic Model of Mobile Robot
DC Motor Model

The angular velocity of the mobile robot with respect to ICC is given as follows:. Interchanging the right and left angular velocities, the total angular velocity of the mobile robot is given as:. The current radius of curvature of the mobile robot path relative to the center of the wheel axis is given as:.

Reducing the block diagram gives the system a second-order transfer function from the input armature voltage to the resulting speed change, which is given as:.

Figure 3.1: Kinematic model of the WMR [23, 52]

Controller

Conventional Controller
Fuzzy Logic Controller

Fuzzy with Three membership function (F3M) Controller. 30
Fuzzy with Seven membership function (F7M) Controller 38
Fuzzy and Proportional (FP) Controller
Fuzzy with Proportional Integral (FPI) Controller
Fuzzy and Proportional Integral Derivative (FPID)

Optimization of Path Planning

Principles of Ant Colony Optimization Technique
Ant Colony Optimization (ACO) Algorithm

The neighboring ant is shown in Figure 3.26. where N is generally the set of all adjacent locations of the current location. The ant takes its next step randomly, based on the probability given by. 3.31) where  is the accumulated pheromone at the jth grid point when the ith grid point is the ant's current location at that time. De is the dot product of the vector from to with the vector i to the destination point.

The first term is related to the amount of pheromone at the 8 neighboring points of the network (a local pheromone). The global term is defined as the dot product of the agent direction and the feed (destination) direction. The purpose of the global term is to direct the agent in the desired direction so that a large number of ants can reach the goal point in a limited number of steps.

However, the pheromone will only be deposited if the ant gets to the destination position in less than a certain number of steps. The ACO algorithm used in this experiment is the Ant System (AS) algorithm proposed by Marco Dorigo [58]. However, as shown in Equation 3.32, a new heuristic equation of the state transition rules was used, which is more suitable with applications of this research.

An accurate distance value by heuristic comparison and the higher amount of pheromone of the visited node will be obtained by the ants having a higher probability of choosing that node. During the construction of the path, the pheromone is locally reduced by the given evaporation rate using the formula of update local rules.

Summary

RESULTS AND DISCUSSION

Controller

Conventional Controller
Fuzzy Logic Controller (F)
Hybrid Controller

Fuzzy Logic with Proportional Controller (FP)
Fuzzy logic with proportional integral controller (FPI)
Fuzzy logic with proportional integral derivative controller

Meanwhile, Figure 4.3 shows the position of the mobile robot with respect to the x-position and y-position frame. It can be seen from Figure 4.3 that the WMR position using conventional controllers for 10x10y is the same for P, PI and PID controllers. The graphs in Figures 4.5, 4.6 and 4.7 show the comparison of the simulation results obtained from the response of the robot with the difference membership function of the fuzzy logic controller.

Meanwhile, Figure 4.7 shows the position of the mobile robot relative to the x and y position of the frame. It can be seen from Figure 4.5 that the angle change occurs until it reaches a steady state at 2.3 seconds for the F3M, F5M and F7M controllers. The graph in Figures 4.9, 4.10 and 4.11 shown below compares the simulation results obtained from the membership function response of the fuzzy logic difference with the proportional robot controller.

Meanwhile, Figure 4.11 shows the position of the mobile robot relative to the x and y position of the frame. Figure 4.9 shows that the angle change occurs until steady state is reached at 3.7 seconds for the F3MP, F7MP and 3.6 seconds for the F7MP controller. From Figure 4.13 it can be seen that the angle change occurs until the steady state is reached after 2.3 seconds for the F3MPI, F5MPI and 2.25 seconds for the F7MPI controller.

From Figure 4.20, the model for differential drive WMRs simulated using different types of fuzzy logic with various membership functions with proportional integral derivative controllers such as F3MPID, F5MPID and F7MPID for different target points gives different values for time (in seconds). The kinematic model of WMR (the final Equation 3.9) and the motor model based on Maxon EC-4 pole 30 DC brushless motor (Equation 3.28) were used as the plant in all the simulation work carried out by the research, it is clearly shown in all the simulation block diagrams (Figure 3.9 for conventional controller, Figure 3.12 for fuzzy logic controller and Figures for the hybrid controller).

Figure 4.1: Comparison of angle of orientation of P, PI and PID controller for 10x10y It can be seen from Figure 4.1 that the change of angle occurs until it reaches steady state at 1.8 seconds for P, PI and PID controllers

Path planning optimization

The plots in Figures 4.23 and 4.24 show the simulation results of the path planning optimization with two obstacles the ant colony optimization (ACO) algorithm. The plots in Figures 4.29 and 4.30 show the simulation results of the path planning optimization with five obstacles of the ant colony optimization (ACO) algorithm. The plots in Figures 4.31 and 4.32 show the simulation results of the path planning optimization with six obstacles the ant colony optimization (ACO) algorithm.

The graphs in Figures 4.33 and 4.34 show the simulation results of the path planning optimization with seven obstacles ant colony optimization (ACO) algorithm. The graphs in Figures 4.37 and 4.38 show the simulation results of the path planning optimization with nine obstacles ant colony optimization (ACO) algorithm. The graphs in Figures 4.39 and 4.40 show the simulation results of the path planning optimization with nine obstacles ant colony optimization (ACO) algorithm.

The path length of five different simulations with different number of obstacles using ACO is given in Table 4.7 and the average path length of five different simulations with different number of obstacles is given in Figure 4.41. As for the length of the path versus the number of obstacles, it showed a non-linear, random decreasing trend. As for the average CPU time with five times of simulation with different number of obstacles, it can be observed that the CPU time decreases with the increase of the number of obstacles.

Regarding path length versus the number of obstacles, the observed trend is that the path appears to decrease as obstacles increase. Regarding the average CPU time versus the number of obstacles, the observed trend is that as the CPU time decreases, the obstacles increase.

Figure 4.21:The best tour found at the 1 obstacle

Summary

CONCLUSIONS

Critical Evaluation of Achievements

The implementation of both questions was discussed in Chapters 3 and 4, where the modeling of the mobile robot was first presented in terms of the WMR kinematic model and the mathematical model of the DC motor. This is followed by a discussion of the simulation work using controllers such as conventional controller (P, PI, PID), fuzzy logic controllers (F3M, F5M and F7M) and the proposed hybrid controller (Fuzzy-PID). Finally, he discussed the optimization of WMR route planning in terms of the principles of ant colony optimization technique and ant colony optimization algorithm.

It can be concluded that the Hybrid controller gives the best results in terms of system performance compared to the PID and Fuzzy Logic controllers. Overall, the Fuzzy logic control with three membership function Proportional Integral controller (F3MPI) gives the better performance compared to the fuzzy logic controller with higher number of membership functions. Moreover, optimization of path planning using ant colony optimization (ACO) technique performed showed that the path length and the CPU time will decrease when the obstacles are increased.

This can be attributed to the fact that the path planning method based on ACO can find a near-optimal path and avoid obstacles in a timely manner, both in simple (fewer obstacles) and complex (many obstacles) environments. The reason for this is that ACO uses the characteristic of the solution space in path planning.

Suggestions for Further Work

Concluding Remarks