| Mobile robot path planning is comprised of On-line global path planning,which based on model that the environment of the robot is certain,and Off-line local path planning,which based on sensor that the environment of the robot is uncertain. Because of characteristics of parallelism and be proficient at global search,genetic algorithm(GA)is applied to robot path planning.But the modeling method in these applications is either Grid method or Visibility graphs,the former has difficulty in deciding the grid's sign,the latter has great dependence on obstacles.In addition,the GA converges to the global best slowly,so the existing path planning method based on GA is inefficient.So,a novelty modeling method is presented for GA in robot path planning,which is implemented only dependent on the positions of the start node and the goal node of the robot,avoiding the disadvantages of Visibility graphs method and Grid method.Genes in this method lie in the vertical lines of nodes that equally divide the line from the start node of the robot to the goal node.To speed up the convergence of GA,the positive feedback in the Ant Colony Optimization(ACO)is introduced to basic GA.To reserve quality sequences of chromosome into next generation,a matrix is used for registering the value of working space.A gene with higher value participates in crossover operation and mutation operation with lower probability.The experiment results demonstrate positive feedback GA algorithm can immediately find a path even in complex environments.To get better path,adding genes operation and deleting genes operation are introduced to positive feedback GA,which can generate genes not only in the vertical lines.Associated with rolling window planning,this algorithm is applied to local path planning.The simulation result shows that this algorithm is effectively not only in uncertain environment without mobile obstacles but also in uncertain environment with mobile obstacles. |
PDF Full Text Request