混合遗传算法在机器人路径规划中的应用

Application of Hybrid Genetic Algorithm in Robot Path Planning

针对传统智能算法在多障碍物环境下求解路径时存在忽视路径安全性,易陷入局部最优解等问题,提出一种融合粒子群算法(PSO)、遗传算法(GA)和人工势场法(APF)的混合遗传算法(PA-GA)。首先,改进障碍物参数和算法的适应度函数,引入防碰撞距离与安全距离,保证路径安全性;其次,通过动态调整粒子群算法中的惯性权重增强粒子的搜索能力,加快算法收敛;然后,引入分群策略、等级交叉策略和人工势场法来改进遗传算法的交叉变异操作,依靠自适应调整交叉变异概率加快收敛速度;最后,将改进后的算法融合,保证混合算法在全局和局部的寻优能力。仿真结果显示,PA-GA算法具备了较强的寻优能力,且路径检索结果更好,收敛速率也更快。

Concerning the existing problems in traditional intelligent algorithm that the path security will be ignored and the algorithm will fall into the local optimal solution easily when solving the path in a multi-obstacle environment,a hybrid genetic algorithm(PA-GA)fusing particle swarm optimization(PSO),genetic algorithm(GA)and artificial potential field(APF)is proposed.Firstly,the obstacle parameters and the fitness function of the algorithm were enhanced,and the anti-collision distance and safety distance were introduced to ensure the safety of the path.Secondly,the algorithm convergence was accelerated by dynamically adjusting the inertia weight in the PSO to enhance the search ability of particles.Then,the grouping strategy,level crossover strategy and APF were introduced to improve the crossover mutation operation of GA,resulting in the acceleration of the convergence speed through adaptively adjusting the crossover mutation probability.Finally,the improved algorithms were fused to ensure the ability of both global and local optimization in the PA-GA.The simulation results show that the PA-GA algorithm has strong optimization ability,better path retrieval results and faster convergence rate.