Format: MS WORD Chapters: 1-5
Pages: 66 Attributes: STANDARD RESEARCH
1.0 BACKGROUND OF THE STUDY
Mobile robots have been successfully applied in many areas such as medical and military applications, ocean and space exploration, sports and entertainment, security, public and domestic duties e.t.c of which their environments can either be static or dynamic. They can perform difficult and hazardous tasks with complex requirements and often have to do so autonomously, without the aid of a human operator. Path planning is a very important task for the autonomous mobile robot. It is necessary to ensure a collision-free motion in an obstacle-prone environment in order to navigate safely from the start state to the goal state. Normally, there are various feasible paths for a robot to reach the target from the start location, but in circumstance, the best feasible path is selected according to some guideline such as shortest distance, smoothness of the path, minimum energy consumption etc. or the most adopted criteria is the shortest distance with the minimal possible time. Collision-free path planning plays an important role in mobile robots navigation and is often a fundamental requirement for proper task execution. This complex task poses many difficulties: computational complexity, adaptation to changing environment and determining a reasonable evaluation function for the generated path.
Mobile robots path planning research field commenced in the middle of 1960s. At the beginning, researchers worked on static environments and used statistical and mathematical methods such as Artificial Potential Field and Visibility Graph to solve the problem. Many efforts have been conducted in robotics research for solving the fundamental problem of motion planning, which consists of generating a collision-free path between start and goal positions for a robot in a static and completely known environment, where there could be obstacles. Mobile robot motion planning in dynamic environments has been studied extensively in. On-line planning algorithms are needed for changing environments, but these typically suffer from a lack of generality in the knowledge of the space in which they are executed. Lots of studies exist on the motion planning for robotic systems using various approaches. There is strong evidence that a complete planner (i.e. one that finds a path whenever one exists and reports that no one exists otherwise) will take time exponential in the number of degrees of freedom (dof) of the robot, and that the algorithm belongs to a class of problem known as NPComplete.
Recently some classical approaches, such as cell decomposition, potential field method, road map and sub goal network have been presented in the field of mobile robotics. In a cell decomposition method a two-dimensional map is divided into several grids and the path is created in them. Another case of a classical approach is a potential field method in which the controlled robot is attracted by the destination while simultaneously being repelled by the obstacles. These path planning algorithms suffer from some drawbacks, e.g., a solution may not be optimal because the algorithm gets stuck in local minima or a new solution has to be generated again when the environment changes and therefore the original path can become infeasible. As a result, many heuristic based methods, such as fuzzy logic, artificial neural network, nature inspired algorithms and hybrid algorithms were created. These methods can overcome drawbacks of the classical ones, but they do not guarantee to find the best solution. Still, the result can be sufficiently close to the optimal one.
Particle Swarm Optimization (PSO) is a metaheuristic algorithm which is inspired by the social foraging behaviour of some animals such as bird flocking and fish schooling. It was developed by Kennedy and Eberhart in 1995. In Particle Swarm Optimization (PSO), a problem space is covered with an initial population of solutions in which they are guided to search for the optimum over a number of generations. The concept of PSO is that each particle randomly searches as through the problem space by updating itself with its own memory and the social information gatherers from other particles. An attractive feature of PSO is that its implementation is simple and effective and if the path exist this algorithm can find it. Movement of a robot position is realized by the Particle Swarm Optimization algorithm. PSO convergence to the best solution by adjusting the trail of each individual particle toward its best location based on the best of itself and global best on the neighbour particles. The modification of a robot position is realized by position and velocity information.
The common problems that motivates researchers in the field of autonomous robot navigation is the ability to ensure these robots move safely and fast from its start state to its goal state avoiding any obstacle along its way especially in an unknown environment and also producing an optimal result. Examples of implementation of autonomous robot navigation algorithm includes automobile industry (self-driving cars), unmanned spacecraft e.t.c.
The limitation of the research work carried out by Memon et al. (2015) and Bilbeisi et al. (2015) are the key motivation for this research work. These include;
The inability to generate safe path for mobile robot from the start state to the goal state in an unknown environment Memon et al. (2015) and also navigating in an unknown environment needs online presence which is the drawback of the research work of Bilbeisi et al. (2015).
1.2 PROJECT OBJECTIVES
The specific objectives of this project are to:
a) Design a Particle Swarm Optimization (PSO) based model for mobile robot path planning.
b) simulate the result of the developed model and;
c) evaluate the performance of the model.
In order to achieve the objectives stated above, the below listed methods will be implemented for this research work:
a) Previous path planning algorithm structure will be reviewed.
b) A PSO based model will be developed to plan the path for the mobile robots.
c) The PSO based model is implemented to determine the optimal route of a robot from source to destination point until any obstacle is detected on its path. The obstacle-avoidance decisions made for the robot is only on past and current obstacle motion data obtained by the vision system. Once any obstacle is detected over the optimized path, the obstacle avoidance is done by moving robot towards the nearest safe point around the obstacles boundary.
d) The population is first initialized with random positions, velocity, target and epsilon. Then, the system evaluates the fitness of each particle using equation (1) which is the distance of robot from target position in a particular direction. If any particle’s current fitness is smaller than previous fitness in any axis, then this system saves the current position of the particle as Pbest, otherwise recall previous best position of particle, where particle achieve less distance from the target. After calculation of personal best values of all particles, compare all particle’s best positions (Pbest) and select one particle that has best position among all and point out that location as Global best (Gbest). Then, PSO algorithm is applied for finding the next velocity and location of each particle using equation (2) and (3), respectively. Large values of velocity may results in wide change of particle’s position which causes the divergence problems and particle may leave the search space. Therefore, to handle these situations, inertia weight and velocity clamping techniques are used to reduce the effect of previous velocity on the next velocity. The formula for Inertia weight “ω” is represented by equation (4).
Fitness Function = |Target − Current Location|....................... (1)
vij t+1 = [vij t × ωt+1] + [c1 × r1j t × [Pbest ,i t − xij t ]]+ [c2 × r2j t × [Gbest − xij t ]]..............(2)
xij t+1 = xij t + vij t+1 .............................................(3)
ωt+1 = ωmax − ωmax −ωmin Total Number of Iterations × Current Iteration........................................(4)
1.4 CONTRIBUTION TO KNOWLEDGE
At the end of this project, an efficient Particle Swarm Optimization (PSO) model for robot navigation would have been developed.
|BANKING AND FINANCE||11|
|CONSTRUCTION AND BIULDING||1|
|ELECTRICAL AND ELECTRONICS||1|
|ENGLISH LITERARY STUDIES||30|
|GEOGRAPHY AND PLANNING||1|
|HOM SCIENCE AND MANAGEMENT||3|
|LIBRARY AND INFORMATION SCIENCE||4|
|OFFICE TECHNOLOGY AND MANAGEMENT||21|
|SCIENCE LABORATORY TECHNOLOGY||19|
|SOIL AND ENVIRONMENTAL SCIENCE||1|
No data found...