水下机器人基于侧线机理对目标感知方法的研究

Research on target-sensing method of underwater robots based on the lateral line mechanism

为实现自主水下机器人(AUV)对流场环境的感知,提高其环境自适应能力,根据鱼类侧线感知机理,对复杂流场环境中水下目标的感知方法进行研究。文章通过计算流体力学(CFD)对水下目标体(圆形、三角形和正方形)在均匀流场中产生的扰动流场进行分析,提取一种流场环境中确定位置处的压力信号作为识别信息数据库,训练并建立支持向量机(SVM)模型。通过采集不同位置和不同流场环境中的流场信号对目标体的种类进行识别,结果表明采用8阶及以上的多项式核函数,可以对不同流场环境中的目标体种类进行有效识别,侧线相对目标体的位置对识别精度影响不大;同时通过侧线压力系数波形中的特征值,对目标体的位置进行定量分析,得到不同目标体的位置定位公式,并对其适用的流场环境范围进行了限定说明。结果证明,所提出的基于流场信息进行水下目标感知的方法具有可行性,为了解侧线作用机理和工程化应用提供了一种新思路。

In order to realize the flow sensing of Autonomous Underwater Vehicles(AUV) and improve their adaptive ability, a method of underwater target sensing in a complex flow environment is studied based on the lateral line sensing mechanism. The perturbation of an underwater target(circle, triangle and square) in a uniform flow field is studied by Computational Fluid Dynamics(CFD). In one kind of flow field, the pressure signal on the lateral line is extracted to be the recognition information database. A Support Vector Machine(SVM) model is trained and established, and the types of targets are identified based on the lateral line signals. The results show that the SVM with an 8-order polynomial kernel function can be used to identify the types of targets in different flow environments. And the detection distance has little influence on the recognition accuracy. At the same time, the position of targets is analyzed quantitatively. The location formulas of different targets are obtained through the characteristic values of lateral line pressure signals, and the applicable range is also defined. Therefore, the results show that the proposed method for flow sensing is feasible. And it provides a new idea for understanding the lateral line mechanism and engineering application in the future.