Research of Bio-Inspired Geomagnetic Navigation for AUV Based on Evolutionary Gradient Search

Jiaojiao Guo,Mengfan Wang,Kun Liu,Mingyong Liu,Yun Niu

doi:10.1051/jnwpu/20193750865

Abstract

At present, bio-inspired geomagnetic navigation is mostly based on evolutionary strategy, which requires long navigation time and low efficiency. To solve this problem, a bio-inspired geomagnetic navigation method for AUV based on evolutionary gradient search is proposed. Combining the bionic evolutionary search algorithm with the classical gradient algorithm to search the function extremum can not only ensure the global optimization of the search, but also have fast convergence, which can improve the efficiency of bio-inspired geomagnetic navigation. The simulation results show that this method does not need prior geomagnetic information and can complete navigation tasks according to the geomagnetic trend. Comparing with the evolutionary search strategy, the effectiveness and superiority of the evolutionary gradient search strategy are verified.

Highlights

西北工业大学学报 Journal of Northwestern Polytechnical University https: / / doi.org / 10.1051 / jnwpu / 20193750865
At present, bio⁃inspired geomagnetic navigation is mostly based on evolutionary strategy, which requires long navigation time and low efficiency
A bio⁃inspired geomagnetic navigation method for AUV based on evolutionary gradient search is proposed

Summary

Introduction

西北工业大学学报 Journal of Northwestern Polytechnical University https: / / doi.org / 10.1051 / jnwpu / 20193750865 目前典型的进化搜索算法有着优秀的全局特性,可以广泛应用于搜索导航中,但是由于载体移动过程中的航向角的选取是依据某一概率随机进行的,使得导航过程中搜索耗时长,效率低。基于梯度下降的搜索方法具有快速收敛性,但是容易陷入局部最优,当优化目标不可解析时,梯度搜索的方法就不可再用。对于多目标搜索算法,文献[8] 中提出一种进化梯度搜索策略 ( evolutionary gradient Search, EGS) ,将梯度搜索具有快速收敛的优点与多目标进化算法的全局优化特性结合起来,利用搜索极值完成目标函数的寻优。本文将进化梯度搜索策略用于 AUV 地磁仿生导航中,弥补了进化搜索中耗时长的缺点,也解决了基于梯度信息的线性搜索方法易陷入局部最优的不足。为约束条件,Sk 为 k - 1 到 k 时刻 AUV 的导航路径, B 进化梯度搜索算法( EGS) 是由 Ralf Salomon 提出的一种将典型梯度下降算法与进化算法结合起来搜索函数极值的方法[10] 。进化梯度搜索算法最初应用于气味源搜索中,机器人可以根据气味的浓度趋势来搜索气味源所在,完成气味源搜索导航任务。地磁仿生导航的本质也是利用地磁趋势来完成导航任务的,将进化梯度搜索算法应用于地磁仿生导航中,利用地磁的趋势性来引导 AUV 向目标位置地磁值不断靠近。

Results

Conclusion