基于改进蝙蝠算法的水下机器人避障路径优化方法  

Optimization Method of Obstacle Avoidance Path for Underwater Robot Based on Improved Bat Algorithm

作　　者：龚成勇[1] 刘康 曾永亮 GONG Chengyong;LIU Kang;ZENG Yongliang(School of Energy and Power Engineering,Lanzhou University of Technology,Lanzhou 730050,China;Institute of Water Ecology and Water Engineering in Yellow River Basin,Lanzhou University of Technology,Lanzhou 730050,China)

机构地区：[1]兰州理工大学能源与动力工程学院,甘肃兰州730050 [2]兰州理工大学黄河流域水生态与水工程研究院,甘肃兰州730050

出　　处：《探测与控制学报》2022年第4期118-122,128,共6页Journal of Detection & Control

基　　金：科技部创新方法工作专项项目(2020IM030400);流体与动力机械教育部重点实验室开放基金项目(LTDL2020-001)。

摘　　要：针对水下机器人避障研究中基本蝙蝠理论过早收敛,容易达到局部极值点的问题,对其进行了改进。改进方向主要是在算法中以线性渐变方式调节响度和脉冲发射率,通过脉冲发射率的大小控制蝙蝠进行全局搜索或局部精确搜索,优化了收敛速度慢的问题,缩短了机器人的路径定位时间。在局部寻优阶段融合高斯柯西变异策略,前期利用柯西函数两翼高概率特性产生广范围随机数,扩大算法搜索范围,后期利用高斯函数中间概率高、两端概率低的特性进行精准定位,保证机器人全局最优路径规划的同时,提高了局部精度。同时采用界限随机重置机制,优化了蝙蝠在边界聚集带来的种群多样性不足问题,提高收敛效率,满足了机器人实际定位有效性。通过数值模拟的方法建立水下三维模型,分别利用改进算法和基本蝙蝠算法进行路径规划,模拟表明,改进后的蝙蝠算法相较于基本蝙蝠算法各项指标得以提升,对水下机器人工作运行的避障能力更强,路径规划平顺且长度短。To improve the problem that the basic bat theory converges prematurely and easily reaches the local extreme point in the obstacle avoidance research of underwater robots,the algorithm was adjustedto the loudness and pulse emission rate in a linear gradual manner.Through the size of the pulse emission rate,the bat was controlled to perform a global search or a local precise search,which optimized the problem of slow convergence and shortens the path positioning time of the robot.In the local optimization stage,the Gaussian Cauchy mutation strategy was integrated.In the early stage,the high probability characteristics of the two wings of the Cauchy function were used to generate a wide range of random numbers to expand the search range of the algorithm.In the later stage,the characteristics of the Gaussian function with high middle probability and low probability at both ends were used for precise positioning.While ensuring the global optimal path planning of the robot,the local accuracy was improved.At the same time,the random reset mechanism of the boundary was adopted to optimize the insufficient population diversity caused by bats gathering at the boundary,improve the convergence efficiency,and meet the actual positioning effectiveness of the robot.The underwater three-dimensional model was established through numerical simulation,and the algorithm in this paper and the basic bat algorithm were used for path planning.The method was testified that the ability to avoid obstacles was stronger,the path planning was smooth and the length was short.

关 键 词：改进蝙蝠算法 水下机器人 高斯柯西变异 界限随机重置 数值模拟 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]