Abstract:Although the crossover and mutation of traditional genetic algorithm are randomization and simple, they can produce infeasible path in path planning, increase amount of computation, have the impact on the convergence speed of algorithm. Aimed at this problem, genetic manipulation based on traditional genetic algorithm is improved by using prior knowledge to guarantee the feasibility of the path after genetic manipulation, at the same time, a new adaptive mode for adjusting genetic parameters is presented and matched. The search efficiency of the algorithm optimization was improved. At last, as it was easy for genetic algorithm to fall into the optimal local, the accept determination was proposed on the new units generated through genetic opperation by using the Metropolis rules based on simulated annealing algorithm. The improved genetic algorithm was compared with other genetic algorithm. The result shows that the improved genetic algorithm has obvoius better convergence speed, searching effects and optimization capabilities.