基于改进人工鱼群算法的移动机器人路径规划问题研究

发布时间：2018-01-23 19:40

本文关键词： 人工鱼群算法移动机器人路径规划差分进化算法无人机　出处：《大连海事大学》2017年硕士论文　论文类型：学位论文

【摘要】：路径规划问题是移动机器人技术的相关研究热点之一。该问题具有复杂性、约束性及非线性的特点,具有一定的求解难度。各种智能算法的相继提出,为机器人路径规划问题的较好求解提供了有效的途径。本文采用群智能算法的求解策略。首先提出一种高效的优化算法。以较为新颖的人工鱼群算法为基础,为改善其易早熟、运行效率低等缺陷,提出了一种基于差分进化的并行自适应人工鱼群算法,其基本思路,一是采用并行计算,将种群分为两个子群体,分别注重全局搜索与局部开发,并采用不同参数的自适应策略,两者定期进行信息交流,以提高群体多样性及算法的收敛速度;二是采用混合策略,引入差分进化,当公告板状态多次未得到更新时,对种群进行差分进化操作,可避免算法陷入局部最优;三是对算法的行为选择方式进行改进,引入判定概率,使种群向全局最优移动的概率随算法运行逐渐增大,这样既减少了算法运算量又提高了算法后期的收敛速度。将所提算法应用于经典函数优化问题中,结果验证了其可行性和有效性,所提算法的收敛速度和寻优精度得到了显著提高。然后,将所提的算法应用于机器人路径规划问题之中。本文在该问题的环境建模方面,提出了一种栅栏线数目的确定方法,同时在路径评价函数中考虑了路径长度、平滑度和安全度等多个性能指标。将所提算法应用于机器人的全局路径规划,有障碍物不定时加入的路径规划和变动目标点的路径规划等多种情况下的问题,所得路径长度、平滑度等指标得到明显改善,运行效率得到极大提高。进一一步,将其应用于同属机器人路径规划范畴的无人机航路规划问题之中,仿真结果表明,采用所提算法能够得到一条避开威胁区域且长度较短的优化航路。研究表明所提算法在整体性能上有较大改进和提高,在机器人路径规划问题的多种工况仿真测试中,均有较好的可行性和有效性。本文工作为相关算法更好地求解路径规划等问题,提供了启发和借鉴,具有一定的理论意义和实用价值。
[Abstract]:Path planning is one of the research hotspots in mobile robot technology. It has the characteristics of complexity, constraint and nonlinearity, and it is difficult to solve. A variety of intelligent algorithms have been proposed one after another. It provides an effective way to solve the robot path planning problem. In this paper, an efficient optimization algorithm is proposed, which is based on a novel artificial fish swarm algorithm. A parallel adaptive artificial fish swarm algorithm based on differential evolution is proposed in order to improve its precocity and low running efficiency. The basic idea is to divide the population into two subpopulations by parallel computation. They pay attention to global search and local development, and adopt adaptive strategy of different parameters, and exchange information regularly between them to improve the diversity of population and the convergence speed of the algorithm. Second, the hybrid strategy is used to introduce differential evolution. When the bulletin board state is not updated many times, the population is operated by differential evolution, which can avoid the algorithm falling into local optimum. The third is to improve the behavior selection method of the algorithm and introduce the decision probability to make the probability of population moving to the global optimum increase gradually with the algorithm running. The proposed algorithm is applied to the classical function optimization problem, and the results verify its feasibility and effectiveness. The convergence speed and optimization accuracy of the proposed algorithm are improved significantly. Then, the proposed algorithm is applied to the robot path planning problem. A method for determining the number of fence lines is proposed, and the path length, smoothness and safety are considered in the path evaluation function. The proposed algorithm is applied to the global path planning of the robot. In many cases, such as path planning with obstacles and path planning of changing target points, the obtained path length, smoothness and other indicators are obviously improved. The operational efficiency has been greatly improved. Further, it is applied to the UAV route planning problem, which belongs to the same category of robot path planning. The simulation results show that. By using the proposed algorithm, an optimized route can be obtained, which avoids the threat area and has a short length. The research shows that the proposed algorithm has great improvement and improvement in the overall performance. In the simulation test of robot path planning problem under various conditions, there is good feasibility and effectiveness. The work in this paper provides inspiration and reference for the related algorithms to solve the path planning problems better. It has certain theoretical significance and practical value.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TP242

【相似文献】