基于场景识别的惯性基类脑导航方法

针对无人机导航中惯性器件产生的漂移误差不断随时间累积进而影响导航精度的问题，以仿生类脑导航为研究背景，提出了一种新的导航方法。与传统导航策略不同的是，该方法从周期性校正累积位置误差的角度出发，采用设置位置细胞节点的思路，利用训练好的卷积神经网络模型在细胞节点处进行图像匹配，从而在位置细胞节点处实现惯导位置漂移误差校正。同时，建立节点之间的漂移误差模型，调整误差方程系数，以达到修正漂移误差的目的。最后，基于无人机飞行实验结果验证了该方法对自主导航的有效性和鲁棒性，该方法能够有效提高无人机导航的精度。

Inertial-based Brain-like Navigation Strategy Based on Scene Recognition

The drift error generated by the inertial devices is accumulating with time, affecting the accuracy in UAV navigation. To solve that problem, a new navigation method based on the bionic brain-like navigation is proposed. Different from the traditional navigation strategy, this method is inspired by setting position cell nodes in bionic brain-like strategy that trains convolutional neural network model to perform image matching at cell nodes, so as to achieve the inertial navigation position drift error correction at the position cell node. In addition, the drift error model between nodes is established and the coefficient of error equation is adjusted to correct the drift error. Finally, based on the flight test results of UAV, the effectiveness and robustness of the proposed method for autonomous navigation are verified. The accuracy of UAV navigation is effectively improved by the proposed method.