Information Technology Reference
In-Depth Information
performance of the risk assessment unit, a computer simulator has been designed to
simulate the direction/trajectory generation module.
2.5
Prediction Module
In terms of scanning, since there is only one active region at a time, the non-active
sectors are dealt with using a prediction module to estimate the position of the
obstacles (particularly the dynamic ones). The prediction module employs a linear
technique to estimate the positions of the obstacles which are not currently being
observed by the vision sensor. The risk assessment unit will continue to use the
estimated position of the obstacles until it receives their actual updated positions
when the system re-scans that particular region.
In the next section, the hardware and software environments used in the proposed
approach are explained followed by results.
3
Experimental and Simulation Environments
The proposed system has been implemented using a Googoltech pan and tilt platform
[4], a Microsoft HD video camera for object detection and identification and an
Acuity laser sensor [9] for accurate range measurements. Associated with these
hardware devices, a wide range of software libraries have been used to control and
analyse the acquired data. The Googoltech C\C++ library has been utilised to control
the pan and tilt. Additionally, the Afrog.net computer vision library has been used to
implement the obstacle detection/identification subsystem.
For testing purpose, a number of maritime scenarios have been created using the
commercially available Virtual Sailor Simulator [10]. Indoor experiments have been
carried out in a visualisation lab at Queen's University Belfast equipped with multiple
projectors and screens to emulate a panoramic view of the maritime environment.
This setup has two major disadvantages. The first issue is the lack of depth
information whereas the other is that the platform remains static which is not
practical. In practice, a laser range sensor will be employed to obtain accurate depth
information. In order to display the obstacles' information to the USV operator or to
the captain of the manned ship, a virtual map is developed using the .Net
environment. This is shown in Fig. 2 where the depth information is obtained from a
previously stored file (in simulations) or from a laser range sensor (in practice).
The following section discusses the results of the experiment and simulation of the
proposed system in detail.
4
Results
The hardware-in-the-loop simulation experiments aim to investigate the performance
of the proposed ODA system in terms of detecting and identifying objects and to
Search WWH ::




Custom Search