In this paper, an active single-object localization system with U-shaped ultrasonic module array is presented. The proposed algorithm in this system has two stages. The first stage is time-of-flight (TOF) estimation of reflected ultrasonic signals. We analyze several distortion cases of reflection signals and provide solutions to improve the performance of TOF estimation. The second stage is TOF-based object localization. A good localization algorithm can suppress the problems caused by inaccurate TOF estimation and improve the accuracy of target positioning. We propose two object-localization algorithms to estimate the object location. One is the unconstrained least-squares method, and the other is the constrained weighted least-squares method based on the eigenvalues-analyzing technique. The simulation results suggest that the performance of the proposed constrained weighted method is better than that of the unconstrained method. In addition, the median smoother and the outlier exclusion scheme are implemented in the overall system to make the object location estimation more stable. The performance of the proposed system is confirmed by the experiment results, which show that the single-object localization has a considerable degree of accuracy and stability.

中文翻译：

---

使用多个超声波传感器和约束加权最小二乘法的单目标定位

在本文中，提出了一种具有 U 形超声模块阵列的主动单目标定位系统。该系统中提出的算法有两个阶段。第一阶段是反射超声信号的飞行时间 (TOF) 估计。我们分析了反射信号的几种失真情况，并提供了提高 TOF 估计性能的解决方案。第二阶段是基于 TOF 的对象定位。一个好的定位算法可以抑制TOF估计不准确带来的问题，提高目标定位的精度。我们提出了两种目标定位算法来估计目标位置。一种是无约束最小二乘法，另一种是基于特征值分析技术的约束加权最小二乘法。仿真结果表明，所提出的约束加权方法的性能优于无约束方法。此外，在整个系统中实施了中值平滑和异常值排除方案，使对象位置估计更加稳定。实验结果证实了所提出系统的性能，表明单目标定位具有相当程度的准确性和稳定性。