Research On Application Of Generic Algorithms For Multi-object In The Robot's Path Planning

Robot technology is a combination of multiple subjects, it involves the computer, artificial intelligence, cybernetics, bionics, information and sensing technology, it is also an important symbol of modern scientific progress. The fields of robot's applications are very broad, such as construction, medical, fire, etc. It is an electronic device that combined with the people's intelligence and sensitivity of the machine and ability to work efficiently.This paper mainly study robot's path planning that have different obstacles in two-dimensional space. We need achieve three objectives:security, the path as smooth as possible, the length of path is the shortest. And using MATLAB 7.0 Genetic Algorithm Toolbox that developed by the University Sheffield.The operations of the genetic algorithm have population initialization, selection, crossover and mutation. In generally, solving problems using genetic algorithm, populations are randomly generate with uncertainty. But in my paper, the initial population is not generated in a random manner; the generation and selection of the population have certain selectivity. Finally, put to use MATLAB simulation platform to achieve the goal of security, the path as smooth as possible, the length of path is shortest. And investigate effect of the genetic parameters on the results of running. The same time, investigate the effectiveness in non-randomly generated population's manners.The methods of generated populations with nonrandom is to take same equal parts of points in the connection between the starting point and the end of point, and to do some equal parts of lines that perpendicular to the line of the connection between the start point and end of the point. Each equal part of lines generated a point; the connection of these points will form a population of individuals. But we need to judge every point whether it exists in the internal obstacles, if it does, undesirable; otherwise desirable. Meanwhile, we also need to determine whether the various sections between the generated points is intersection with the edge of the obstacle, if it is, walking along the edge of the obstacle. So each path is feasible and it can ensure safety. If the path of two adjacent inflection points is too sharp, these two paths can be randomly generated points and add nodes to the way to satisfy the smooth performance. So using the methods of generated populations with nonrandom is also preliminarily achieve to security and smoothness goals.Then, distribute different weights to the three objectives whit the weight coefficient method. Sum all of the segments of path as the shortest path; to calculate the angle between the two adjacent paths as smooth performance, it can be solved by cosine theorem, the mean value between the two adjacent paths is more smaller, the mean value of the angle is more larger and that is what we wish; Safety performance requirements the path segment between each vertex and obstacles the farther the better, that is, the distance of point to the path the bigger the better, because all the questions are all minimization problems, so take the countdown of the distance of point to the path as the goal. Finally, use MATLAB platform to do simulation experiments, and draw valid conclusions.