Underwater vehicle obstacle avoidance and path planning using amulti-beam forward looking sonar

This paper describes a new framework for segmentation of sonar images, tracking of underwater objects and motion estimation. This framework is applied to the design of an obstacle avoidance and path planning system for underwater vehicles based on a multi-beam forward looking sonar sensor. The real-time data flow (acoustic images) at the input of the system is first segmented and relevant features are extracted. We also take advantage of the real-time data stream to track the obstacles in following frames to obtain their dynamic characteristics. This allows us to optimize the preprocessing phases in segmenting only the relevant part of the images. Once the static (size and shape) as well as dynamic characteristics (velocity, acceleration,…) of the obstacles have been computed, we create a representation of the vehicle's workspace based on these features. This representation uses constructive solid geometry (CSG) to create a convex set of obstacles defining the workspace. The tracking takes also into account obstacles which are no longer in the field of view of the sonar in the path planning phase. A well-proven nonlinear search (sequential quadratic programming) is then employed, where obstacles are expressed as constraints in the search space. This approach is less affected by local minima than classical methods using potential fields. The proposed system is not only capable of obstacle avoidance but also of path planning in complex environments which include fast moving obstacles. Results obtained on real sonar data are shown and discussed. Possible applications to sonar servoing and real-time motion estimation are also discussed     

Object detection and tracking method of AUV based on acoustic vision

This paper describes a new framework for object detection and tracking of AUV including underwater acoustic data interpolation, underwater acoustic images segmentation and underwater objects tracking. This framework is applied to the design of vision-based method for AUV based on the forward looking sonar sensor. First, the real-time data flow (underwater acoustic images) is pre-processed to form the whole underwater acoustic image, and the relevant position information of objects is extracted and determined. An improved method of double threshold segmentation is proposed to resolve the problem that the threshold cannot be adjusted adaptively in the traditional method. Second, a representation of region information is created in light of the Gaussian particle filter. The weighted integration strategy combining the area and invariant moment is proposed to perfect the weight of particles and to enhance the tracking robustness. Results obtained on the real acoustic vision platform of AUV during sea trials are displayed and discussed. They show that the proposed method can detect and track the moving objects underwater online, and it is effective and robust.     

Curved shape reconstruction using multiple hypothesis tracking

Panoramic sweeps produced by a scanning range sensor often defy interpretation using conventional line-of-sight models, particularly when the environment contains curved, specularly reflective surfaces. Combining multiple scans from different vantage points provides geometric constraints necessary to solve this problem, but not without introducing new difficulties. Existing multiple scan implementations, for the most part, ignore the data correspondence issue. The multiple hypothesis tracking (MHT) algorithm explicitly deals with data correspondence. Given canonical observations extracted from raw scans, the MHT applies multiple behavior models to explain their evolution from one scan to the next. This technique identifies different topological features in the world to which it assigns the corresponding measurements. We apply the algorithm to real sonar scans generated specifically for this investigation. The experiments consist of interrogating a variety of two-dimensional prismatic objects, standing on end in a 1.2-m-deep freshwater tank, from multiple vantage points using a 1.25 MHz profiling sonar system. The results reflect the validity of the algorithm under the initial assumptions and its gradual performance degradation when these assumptions fail to characterize the environment adequately. We close with recommendations that detail extending the approach to handle more natural underwater settings     

Joint direction-of-arrival and array-shape tracking for multiplemoving targets

The problem of tracking the directions-of-arrival (DOAs) of multiple moving sonar targets with an array of passive sensors is complicated by sensor movement. An algorithm for the joint tracking of source DOAs and sensor positions is presented to address this problem. Initial maximum-likelihood estimates of source DOAs and sensor positions are refined by Kalman filtering. Spatio-temporally correlated array movement is considered. Source angle dynamics are used to achieve correct data association. The new technique is capable of performing well for the difficult cases of sources that cross in angle as well as for fully coherent sources. Computer simulations show that the approach is robust in the presence of array motion modeling uncertainty and effectively reduces dependence on expensive and possibly unreliable hardware     

可变尺度区域拟合模型的侧扫声纳分割方法

为解决现有侧扫声纳图像目标分割准确度不高的问题,提出一种联合最大熵去噪和可变尺度区域拟合模型的侧扫声纳图像分割方法。首先,计算图像一维熵,基于最大熵原则对侧扫图像进行降噪处理,提高图像质量,并根据峰值信噪比评判降噪效果;然后基于可变尺度区域拟合模型,采用高斯核函数对分割活动轮廓进行约束,分割降噪后的侧扫声纳图像。通过对含有不同目标物的侧扫声纳图像进行分割实验,验证了联合最大熵去噪和可变尺度区域拟合模型的有效性。     

多波束与侧扫声纳海底目标探测的比较分析

侧扫声纳是目前常用的海底目标(如沉船、水雷、管线等)探测工具,在测深领域,多波束以全覆盖和高效率证明了它的优越性。由于多波束具有很高的分辨率,目前在工程上已经开始应用多波束进行海底目标物的探测。对多波束和侧扫声纳进行了比较分析,并着重探讨了影响多波束分辨率的各种因素。结果表明:多波束的最大优点在于定位精度高,但其适用范围不如侧扫声纳广泛,尤其受到水深和波束角的限制,多波束和侧扫声纳在探测海底目标时具有很好的互补性,同时应用可以提高目标解译的准确性。     

Pulse-length-tolerant features and detectors for sector-scan sonar imagery

This paper presents a neural-network-based system to detect small man-made objects in sequences of sector-scan sonar images created using signals of various pulse lengths. The detection of such objects is considered out to ranges of 150 m by using an experimental sector-scan sonar system mounted on a vessel. The sonar system considered in this investigation has three modes of operation to create images over ranges of 200, 400, and 800 m from the vessel using acoustic pulses of a different duration for each mode. After an initial cleaning operation performed by compensating for the motion of the vessel, the imagery is segmented to extract objects for analysis. A set of 31 features extracted from each object is examined. These features consist of basic object size and contrast features, shape moment-based features, moment invariants, and features extracted from the second-order histogram of each object. Optimal sets of 15 features are then selected for each mode and over all modes using sequential forward selection (SFS) and sequential backward selection (SBS). These features are then used to train neural networks to detect man-made objects in each sonar mode. By the addition of a feature describing the sonar's mode of operation, a neural network is trained to detect man-made objects in any of the three sonar modes. The multimode detector is shown to perform very well when compared with detectors trained specifically for each sonar mode setting. The proposed detector is also shown to perform well when compared to a number of statistical detectors based on the same set of features. The proposed detector achieves a 92.4% probability of detection at a mean false-alarm rate of 10 per image, averaged over all sonar mode settings.     

Geometric distortions in side-scan sonar images: a procedure fortheir estimation and correction

A procedure is introduced for the estimation and correction of geometric distortions frequently observed in side-scan sonar images as a result of motion instabilities of the sonar towfish. This procedure estimates geometric distortions from the image itself, without requiring navigational or altitude measurements. Estimates of the local degree of geometric distortion are obtained by cross-correlating segments of adjacent lines of the image. A mathematical model for the distortions is derived from the geometry of the problem and is applied to these estimates to reconstruct the sampling pattern on the seabed, under the assumption of a planar bottom. The estimated sampling pattern is then used for resampling the image to correct the geometric distortions. The model parameters may also be used for calculating approximate estimates of the attitude parameters of the towfish. A simulation is employed to evaluate the effectiveness of this technique and examples of its application to high-resolution side-scan sonar images are provided     

Single-ping target speed and course estimation using a bistatic sonar

This paper focuses on estimating the two-dimensional (2-D) target-speed vector (course and speed) using a multistatic sonar that consists of one monostatic sonar and one bistatic receiver. The speed and course estimates are obtained after a single transmission. The theory on bistatic Doppler and 2-D target-speed vector estimation is first considered and then applied to simulated and real data. The results can be used to improve classification algorithms or to feed speed information to tracking algorithms, for example.     

A comparison of inter-frame feature measures for robust objectclassification in sector scan sonar image sequences

This paper presents an investigation of the robustness of an inter-frame feature measure classifier for underwater sector scan sonar image sequences. In the initial stages the images are of either divers or remotely operated vehicles (ROV's). The inter-frame feature measures are derived from sequences of sonar scans to characterize the behavior of the objects over time. The classifier has been shown to produce error rates of 0%-2% using real nonnoisy images. The investigation looks at the robustness of the classifier with increased noise conditions and changes in the filtering of the images. It also identifies a set of features that are less susceptible to increased noise conditions and changes in the image filters. These features are the mean variance, and the variance of the rate of change in time of the intra-frame feature measures area, perimeter, compactness, maximum dimension and the first and second invariant moments of the objects. It is shown how the performance of the classifier can be improved. Success rates of up to 100% were obtained for a classifier trained under normal noise conditions, signal-to-noise ratio (SNR) around 9.5 dB, and a noisy test sequence of SNR 7.6 dB     

Towfish orientation and position estimation

The towfish location and orientation problems that arise in using side-scan sonar to detect objects on the sea bottom are treated separately. Data which locate the towfish relative to the ship are usually deteriorated by multipath receptions and other effects. In order to overcome this serious degradation in the location measurements, a modified Kalman filter is proposed. An estimate of the state transition matrix for this filter is derived, and a means of switching between two Kalman gains is suggested. The feasibility of the proposed filter is justified by a case study. Improved estimates of towfish pitch and heading measurements are obtained by a separate system employing model identification and subsequent Kalman filtering. Application of these methods to data from similar towed side-scan sonar systems should yield significant gains in object location accuracy     

Experimental Evaluation of Tracking Algorithms Used for the Determination of Fish Behavioral Statistics

The fine-scale swimming behavior of fish can now be studied because of the development of sophisticated measurement devices such as multibeam sonar and stereo video systems. However, even with these sensors, improved methods are still required to generate quality estimates of swimming speeds and turn rates. Biologists have commonly relied on pointwise differentiation of noisy position measurements while engineers have focused on Bayesian algorithms to track underwater vehicles. A comparative evaluation of the performance of these tracking algorithms for the analysis of fine-scale behavior of fish was performed using a data set of 100 fish tracks recorded simultaneously with a multibeam sonar and a stereo video camera system. The segmenting track identifier, a non-Bayesian curve fitting and segmenting tracker, is shown to be most effective for tracking the unpredictable and complex horizontal motion of fish while a Kalman smoother using a constant-velocity model is shown to be most effective for tracking the more predictable and piecewise linear vertical motion of fish. Both are shown to be more effective than pointwise differentiation. Criteria for selecting an appropriate algorithm for a given motion study are provided     

基于MRF场的侧扫声呐图像分割方法

为了利用侧扫声呐进行水下目标自动探测和识别,首先必须将声呐图像分为目标高亮区、海底混响区和目标阴影区.由于声呐图像有强背景噪声,传统的图像分割方法显得无能为力,故采用基于MRF场的图像分割方法来准确地分割.根据侧扫声呐目标的成像特点,建立了分割的约束条件;利用阴影与目标的灰度均值比很小这一特点进行初始分割,然后根据分割后目标与阴影的宽度差来剔除虚假目标,由初始分割的结果求得MRF模型初始参数,再采用迭代条件估计得到最终的模型参数和准确的分割结果.由于考虑了相邻像素间的依赖关系,具有抗噪性强、分割效果好的优点,从理论上说是合理的.实测数据分析也证明了这种算法的优越性.     

Position Correction Using Echoes From a Navigation Fix for Synthetic Aperture Sonar Imaging

In synthetic aperture sonar (SAS), the platform position must be known sufficiently accurately for signals to be added coherently along the synthetic aperture. Often, the onboard navigation system is insufficiently accurate by itself, so corrections are needed. A well-known method is the displaced phase center antenna (DPCA) procedure for correcting platform position using seabed echoes. DPCA methods have the advantage of insensitivity to changing interference patterns, moving specular reflection, and changing occlusion, with aspect. However, when seabed echoes are unusable, either because they are too weak, or because they are corrupted by multipath, the seabed DCPA method may fail. Therefore, we present an alternative DPCA method using sonar echoes from a suitable navigation fix, based on an object detected after standard beamforming. In our proposed system, look angle is obtained by tracking the centroid of the rectified image of the fix object. When the standard DPCA correction equations are modified for a fixed reflector, it turns out that they provide incremental range and look-angle errors, precisely the values required when the target itself is used as the navigation fix. Moreover, the values obtained are then self-compensating for errors in estimating seabed depth or forward motion of the platform. The navigation fix is selected by bracketing in range, and beamforming overlapping subsets of the receiver array. In this paper, we present experimental results at transmitter frequencies of 25 and 100 kHz where our method enabled well-focused SAS images to be generated with little recourse to other navigation information. Hence, SAS can be carried out, even when a sophisticated inertial navigation system (INS) is not available.      

基于声呐图像的深海采矿车组合定位导航算法

实现高精度的定位导航是深海采矿车完成海底工作任务的基础条件。在采矿车行进过程中,声呐设备生成的图像信息能够反映海底场景的变化,从而体现采矿车本身的运动,由此建立了一种声呐图像里程计,并将其与轮式里程计和USBL测量数据相结合提出了一种深海采矿车组合定位导航算法。首先对多波束前视声呐图像进行预处理,然后使用Canny算法进行特征检测并对特征点云进行配准,再结合声呐成像原理构建了声呐图像里程计运动模型,最后通过轮式里程计运动模型推导预测方程、声呐图像里程计运动模型和USBL测量数据推导更新方程,利用EKF（extended Kalman filter）算法实现基于多传感器融合的定位与姿态估计。海试数据验证了该组合定位算法能实现轮式里程计、声呐里程计和超短基线在速度、位置、艏向角估计、定位速率的精度互补,具有一定的有效性和精确性,该算法为深海采矿车的定位与导航算法研发提供了参考。     

Direct displacement measurements of submerged objects

The direct measurement of displacement time histories, obtained by tracking illuminated targets located on the exterior surface of submerged objects, has significant advantages for large-scale model tests. Many problems of interest require the simultaneous measurement of three-dimensional motions at several elevations on the same body or on several bodies in close proximity. Thus, optical tracking of targets in air or water or some combination of the two required the development of a general analytical formulation capable of converting local data measured from multiple camera locations to a global coordinate system for further analysis. The formulation presented is quite general and allows one to study the displacement behavior of single or multiple targets located on the same object or on different moving objects. As part of the analysis, calibration issues, possible sources of error and the optimal choice of camera location are discussed in detail. Data obtained from a large-scale model test of deepwater platform TLP tendons subjected to random design seas is used to illustrate the methodology. In the example, four different cameras were used to record the tendon displacements. To illustrate the methodology, five different camera combinations were used to compute the target location. The results indicate that the formulation is quite robust and that, although only a two-camera system is required to resolve the three-dimensional global coordinates of a single target, the use of a third camera can enhance the accuracy of the results. Guidelines for using cameras to track target motion were developed and are presented.     

大深度近海底精细地形地貌探测技术

船载低频多波束测深声纳、侧扫声纳可以对深海海底地形地貌进行快速、高效、大面积探测,但其测量精度有限,难以满足深海科学考察、资源勘探开发对高精度海底地形地貌的需求。随着各类大深度水下移动载体(如深海拖体、水下机器人、遥控潜器和载人潜水器)的涌现,特别是各类耐高压测绘声纳的商业化,使大深度近海底精细地形地貌探测成为可能。首先分析了多波束测深声纳、侧扫声纳和测深侧扫声纳等3种测绘声纳的基本原理,然后分别介绍了各类测绘声纳的国内外典型商业化产品,并通过典型实例分析了其在大深度近海底精细测绘中的应用情况。     

Biomimetic sonar locates and recognizes objects

An active sonar is described that adaptively changes its location and configuration in response to the echoes it observes in order to locate an object, position it at a known location, and identify it using features extracted from the echoes. The sonar consists of a center transmitter flanked by two receivers that can rotate and is positioned at the end of a robot arm that has five degree-of-freedom mobility. The sonar operates in air using Polaroid transducers that are resonant at 60 kHz with a nominal wavelength equal to 6 mm. The emitted pulse has a short duration with a useful bandwidth extending from 20 to 130 kHz. Using binaural information, the transmitter rotates to position an echo-producing object on its axis to maximize the acoustic intensity incident on the nearest echo-producing feature. The receivers rotate to maximize the echo amplitude and bandwidth. These optimizations are useful for differentiating objects. The system recognizes a collection of ball bearings, machine washers, and rubber O-rings of different sizes ranging from 0.45 to 2.54 cm, some differing by less than 1 mm in diameter. Learning is accomplished by extracting vectors of 32 echo envelope values acquired during a scan in elevation and forming a data base. Recognition is accomplished by comparing a single observed echo vector with the data base to find the least squared error match. A bent-wire paper clip illustrates the recognition of an asymmetric pose-dependent object