全天域仿生复眼偏振光自主导航机理研究

Imitation of natural insect compound eye polarization autonomous navigation in recent years has been paid more and more attention, thus has deeply need for further research. This project proposed to use Monte Carlo stochastic process for simulation of photon propagation process and establish precise characterization model for polarized light and reflect of sky depolarization caused by cloudy atmosphere. By setting polarization degree as data clustering factor, the temporal relationship model between the sun and the zenith is established and analysed, followed by derivation and solving method of attitude transformation. Based on that, refer to compound eye visual navigation principle of bees, a novel mechanism and method to reduce the optical flow accumulated position error is proposed and designed by making full use of the compound eye polarized optical flow information and performing iterative Kalman filter combined with optical flow and the predicted optical flow, the parameters and precision criterion are given and the mechanism of Omnimax bionic polarization autonomous navigation is revealed. This project can further improve the bionic compound eye navigation theoretical model and thus has very important significance for scientific research of autonomous navigation in areas such as defense and military.

师从自然的仿昆虫复眼偏振光自主导航方法近几年受到了越来越多的重视，迫切需要进行深入的研究。本项目针对现有方法通常考虑晴朗天气下的理想大气偏振模式采用Rayleigh散射仿真模型不能准确表征混浊大气下云层散射对天空偏振光的反射和退偏的不足，提出引入统计学中Monte Carlo随机过程模拟光子传播过程，建立混浊大气偏振模式的精确表征模型；提出将偏振度作为数据聚类依据，分析和建立太阳和天顶固定时空关系模型，推导姿态变换关系和精确求解载体空间姿态角的方法；在此基础上，借鉴蜜蜂复眼视觉导航原理，提出并设计一种充分利用复眼偏振光信息将光流法和卡尔曼滤波器结合执行实际光流和预测光流的迭代匹配，修正路径积分的累积位置误差的机理和方法，给出实际设计参数和精度判据，揭示全天域仿生偏振光自主导航机理。本项目的研究进一步完善了仿生复眼导航理论模型，对国防军事等领域未来的自主导航具有非常重要的科学研究意义。

本项目针对现有方法通常考虑晴朗天气下的理想大气偏振模式采用Rayleigh散射仿真模型不能准确表征混浊大气下云层散射对天空偏振光的反射和退偏的不足，提出引入统计学中Monte Carlo随机过程模拟光子传播过程，建立了混浊大气偏振模式的精确表征模型；提出将偏振度作为数据聚类依据，分析和建立太阳和天顶固定时空关系模型，推导姿态变换关系和精确求解载体空间姿态角的方法；提出了一种基于转动惯量的航向角解算算法，通过对称轴检查有效获取了载体的航向信息，同时亦提出了一种基于软边缘支持向量机仿生偏振光航向解算方法，利用机器学习中的支持向量机方法抑制云、阴天等恶劣天气的干扰，同时将光流法和卡尔曼滤波器结合执行实际光流和预测光流的迭代匹配，进行成像稳定和位置预测，给出了实际设计参数和精度判据，构建了试验演示系统，揭示了全天域仿生偏振光自主导航机理。本项目的研究进一步完善了仿生复眼导航理论模型，对国防军事等领域未来的自主导航具有非常重要的科学研究意义。