Sea Technology

NOV 2013

The industry's recognized authority for design, engineering and application of equipment and services in the global ocean community

Issue link:

Contents of this Issue


Page 40 of 75

targets (to the vehicular frame) can be streamed continuously to the control system, allowing more complex tasks to be undertaken, i.e., following the track of a riser, surveying around a target or maintaining a constant offset from a chosen target, e.g., a blowout preventer, while monitoring or intervention work is completed. Tracking Objects in Sonar Data Detecting and tracking objects in sonar data is a complex task. The process for acquiring, storing and displaying sonar data is obviously different from common optical imagery. The data usually present different types of environmental noise, and therefore the signal return from an object will vary as the object is visualized in different environments. Detecting known objects can signifcantly simplify the problem, as the developer of the software can use information from the object to help design the algorithms used to fnd those objects. However, a tool that can be used by the entire ROV industry must be able to provide a means for detecting generic and widely different objects. An effective solution is to look for anomalies within the sensor data. Simple flters looking for local differences can be used to process the sonar image and highlight possible areas of interest. This is a very fast process that can be run using images made of integral numbers. The result is an image where regions that do not statistically match the surroundings are fagged as potential objects. By using simple constraints (such as considering the size of certain objects) the user can very quickly fnd objects of interest. This initial detection step is useful in presenting the user with objects that can be tracked and used to assist the control of the ROV. In order for the ROV control to succeed, the ROV user must be able to select an object and then track it through a sequence of frames. Kalman flters and particle flters are both examples of tools that can be used to predict the position of objects on the sonar image, based on previous images. Any tracking algorithm should be able to estimate the position of the object based on the previous observations of that object and also offer a quality fgure associated with the tracking performance. For this application, only one object needs to be tracked, but in order for that object to be selected it is necessary for the algorithms to track multiple objects simultaneously. Therefore, data association algorithms are required. These compare features from the objects and match the correct objects to previous observations. In challenging environments, the ROVs or the targets may be moving fast and unpredictably, or both the ROV and targets may move at the same time. This is a challenge for successful detection and tracking, but, if available, the user can use the information on the ROV position and orientation to improve the prediction from the flters. Such instances of tracking have been used successfully to help with the detection and tracking of targets seen with an MBIS from a VideoRay (Pottstown, Pennsylvania) mini-ROV or an SMD (Wallsend, England) work-class ROV. The VideoRay and SMD solutions are offered by both manufacturers as a standard option to all their customers. In a recent operation, the VideoRay solution has been used successfully to help detect and track a diver. Dynamic Positioning True dynamic positioning (DP) is used to defne the process of automatically controlling a system through the use of WORLD CLASS Through People, Technology & Dedication For more information about our products, please contact us at +1 508-563-6565 November 2013 / st 41

Articles in this issue

Links on this page

Archives of this issue

view archives of Sea Technology - NOV 2013