Center for Coastline Security Technology



[pic]

Center for Coastline Security Technology

Year 1: Final Technical Report

Contract/PR No. N00014-05-C-0031

Prepared for

US Office of Naval Research

For the Period

08 June 2005 to 07 June 2006

Submitted by

Stewart Glegg, William Glenn, Borko Furht, Carl Berkowitz, P. Beaujean, G. Frisk, S. Schock, K. vonEllenrider, P. Ananthakrishnan, Edgar An, R. Granata, R. Coulson

College of Engineering

Florida Atlantic University,

777 Glades Road

Boca Raton, FL 33431

561 297 3000

Submitted June 7th, 2006

ABSTRACT

The Center for Coastline Security Technology (CCST) focuses on research, simulation, and evaluation of coastal defense and marine domain awareness equipment, sensors and components. It builds upon the existing efforts and expertise in coastal systems and sensor research at the Institute for Ocean and Systems Engineering (IOSE), the Imaging Technology Center, the Department of Computer Science and Engineering and the University Consortium for Intermodal Transportation Safety and Security at Florida Atlantic University.

New technologies are needed to enhance surveillance and inspections of marine activities in the coastal zone that includes major ports, small inlets, beaches, remote coastal areas and their approaches. To be efficient and cost effective it is imperative to mount the surveillance systems and sensors on autonomous platforms that can operate unsupervised for extended periods of time. The task is to effectively integrate sensors with underwater, surface and airborne autonomous and remotely operated platforms and to incorporate video and image analysis and data mining methods to quickly and effectively identify threat events. These algorithms will automatically detect moving objects of interest, identify their location, track them through a video sequence, and classify them into semantic categories. In addition this group has unique resources and state-of-the-art capabilities in the area of real time 3D/4D virtual simulation that can be used to evaluate coastline security technologies and systems prior to deployment, and to carry out threat analysis using simulation for both training and exercises.

The three technologies that are being developed in this program are:

1) Underwater vehicles for survey and inspection: The faculty and staff in Florida Atlantic University’s Department of Ocean Engineering has extensive experience in developing, building and operating Autonomous Underwater Vehicles (AUVs). In the proposed program we will develop a low cost, one man operated, remotely piloted unmanned, untethered, underwater vehicle, which will provide real time underwater video and sonar images to a topside console. The specific application to be addressed is underwater inspections by rapid response teams, and routine inspection activities, currently carried out by scuba divers. This technology is intended to reduce the need for divers on a 24/7 basis.

2) High Definition Video Systems: High definition video cameras provide an order of magnitude improvement in field of view and/or range over and above conventional systems. Consequently they are a necessity for harbor surveillance, but their implementation in this environment is limited by size and cost. At Florida Atlantic Universities Imaging Technology Center, a compact high definition camera has been developed and is ready for the commercial market, the primary customers being the film industry. For the port security application there are several research issues that still need to be addressed, specifically, managing the high data output rate of the camera, and testing the camera in the marine environment. The test and evaluation issue has been addressed by the ITC center in collaboration with NAVSEA Carderock’s South Florida Test Facility, which has towers overlooking Port Everglades and the adjacent inlet that are already used by the USCG for video surveillance. The data management issue is being addressed by Florida Atlantic University’s Department of Computer Science and Engineering.

3) Simulation of Port Security Scenarios: For any port, the evaluation of threat scenarios, and the operational procedures required to implement appropriate countermeasures, cannot be fully enacted during training exercises. Further, the effectiveness of surveillance operations needs to be fully evaluated prior to installation of permanent equipment. This is best achieved through computer simulation of the port environment. Florida Atlantic Universities Center for Intermodal Transportation Safety and Security has recognized expertise and software available for this activity, and has initiated proof of concept studies at Port Everglades. In this program a simulation of the activities in Port Everglades has been developed, including the interaction between Navy assets and civilian activities.

This document is the final report for year one of this three year program and describes the progress on the following projects

• The Development of a Remotely Piloted, Unmanned, Untethered, Underwater Vehicle (RPUUV),

• High Definition High Frame-Rate Color Camera for Surveillance

• Video Analysis and Image and Video Data Mining

• Port Everglades Simulation

CONTENTS

Abstract

List of Figures

List of Tables

Executive Summary

1.0 INTRODUCTION

1.1 Overview

1.1.1 Background

1.1.2 Technical Objectives

2.0 THE REMOTELY PILOTED, UNMANNED, UNTETHERED,

UNDERWATER VEHICLE (RPUUV)

2.1 Background

2.2 Development of a Remotely Piloted Unmanned

Underwater Vehicle PI: Dr. Stewart Glegg

2.2.1 Summary

2.2.2 Introduction

2.2.3 System Concept

2.2.4 Overall Vehicle Design

2.2.5 Nose Section

2.2.6 Mid Section

2.2.7 Tail Section

2.2.8 Electronic Components

2.2.9 Tow Float

2.2.10 In Water Test Results

2.3 Acoustic and RF Communications:PI: Dr. P.Beaujean

2.3.1 Summary

2.3.2 Introduction

2.3.3 Remote piloting and Control using Electro-Magnetic Waves

2.3.4 Acoustic remote piloting and positioning

2.3.5 High-speed acoustic communications

References for Section 2.3

2.4 Environmental Assessment and Modeling:

PI: Dr. George V. Frisk

2.4.1 Summary

2.4.2 Introduction

2.4.3 Known Information

2.4.4 Proposed Measurement Program:

2.4.5 Results to date

2.4.6 Conclusion / Discussion

2.5 Collision Avoidance Sonar: PI: Dr. Steven Schock

2.5.1 Introduction

2.5.2 Deliverables

2.5.2.1 Acoustic Projector Model

2.5.2.2 Sonar Design

2.5.2.3 Test Results

2.5.3 Conclusions

2.6 Hydrodynamic design and integration of RP UUV

hull-form and control surfaces: PI: Dr. von Ellenrieder

2.6.1 Summary

2.6.2. Introduction

2.6.3. Propeller and rotor duct design

2.6.3-1. Estimation of propeller thrust and torque

2.6.3-2. Recommendations for propeller and duct profile

2.6.4. Experimental Setup

2.6.4-1. Sting

2.6.4-2. Towing Carriage

2.6.4-3. AUV Model

2.6.4-4. PIV system

2.6.5. Experimental Results

2.6.5-1. RPUUV Aft Flow

2.6.6. Conclusions

References for Section 2.6

2.7 Hydrodynamics Analysis and Simulations for Design

and Operation of Remotely Piloted UUV

PI: P. Ananthakrishnan

2.7.1. Introduction

2.7.2. Estimation of Drag and Propeller Thrust and Power

2.7.3. RPUUV Dynamics - Simulation of Horizontal Plane Motion

2.7.3-1. 6DOF Rigid Body Motion

2.7.3-2. 3DOF Horizontal Plane Motion

2.7.3-3. External Force and Moment

2.7.3-4. Numerical Solution of Equations of Motion

2.7.3-5. Estimation of Hydrodynamic Coefficients

2.7.3-6. Simulations to Determine Effectiveness of Vector Thruster

2.7.3-7. Simulations to Determine Fixed Aft Fins Requirements

2.7.3-8. Recommendation on Aft Fins Requirements

2.7.4. RPUUV Dynamics - Simulation of Vertical Plane Motion

2.7.4-1. Formulation of Vertical Plane Motion

2.7.4-2. External Force and Moment

2.7.4-3. Solution Method

2.7.4-4. Simulations of Vertical Plane Motion

2.7.4-5. Determination of Optimul Location on RPUUV for Tow Float Connection

2.7.5. RPUUV Hydrodynamics - Simulation of Aft Flow Including Propulsion Effect

2.7.5-1. Formulation of Viscous Flow Problem

2.7.5-2. Solution Method

2.7.5-3. Results

2.7.6. Conclusions

2.7 Appendix (listing source Fortran codes)

References for Section 2.7

2.8 RP UUV Navigation and Control: PI: Dr. Edgar An

2.8.1 Summary

2.8.2 Introduction

2.8.3 Methods, Assumptions, and Procedures

2.8.3.1 Environment

2.8.3.2 Vehicle

2.8.3.3 Ray-tracing

2.8.3.4 Acoustics

2.8.3.5 Pressure Wave Calculations

2.8.3.6 Obstacle Avoidance System (OAS)

2.8.4 Results and Discussion

2.8.4.1 Simple Cylinder

2.8.4.2 Test Tank Following

2.8.4.3 Half Moon Canyon

2.8.5 Conclusion

2.8.6 Recommendations

2.8.7 References

2.9 Chemical Sensors: PI: Dr. Richard Granata

2.9.1 Summary

2.9.2 Introduction

2.9.3 Methods, Assumptions, and Procedures

2.9.3.1 Primary Test Materials

2.9.3.2 Primary Test Equipment

2.9.3.3 Experiments

2.9.4 Results and Discussion

2.9.5 Conclusions and Recommendations

References for Section 2.9 - Chemical Detector

3.0 HIGH DEFINITION HIGH FRAME-RATE COLOR CAMERA FOR

SURVEILLANCE : PI: Dr. William Glenn

3.1 Summary

3.2 Objective

3.3 Results and Discussion

3.4 Patents Filed

3.5 Conclusion

3.6 Recommendation

4.0 VIDEO ANALYSIS AND IMAGE AND VIDEO DATA MINING

PI: Dr. Borko Furht

4.1. Summary

4.2 Introduction

4.2.1. Project Description

4.2.2. Project Scope and Objectives

4.2.3. Overview of Visual Surveillance

4.2.4. Project Team

4.3 Video Analysis for Object Detection and Extraction

4.3.1. Related Work

4.3.2. Description of the Proposed Algorithm

4.3.3. Results

4.4. Object Classification

4.4.1.1. Literature Survey on Object Classification

4.4.1.2. Proposed Approach

4.4.2. Detection of Ships

4.4.3. Feature Extraction

4.4.4. Classification Algorithm

4.4.4.1. k-Nearest-Neighbor Algorithm

4.4.4.2. Artificial Neural Network

4.4.5. Performance Evaluation by Cross Validation

4.4.6. Experimental Results

4.4.6.1. k-Nearest-Neighbor Algorithm

4.4.6.2. Artificial Neural Network

4.4..6.3. Comparison of Performance

4.5 Conclusions and Recommendations

References·

5.0 VIRTUAL PORT SIMULATION-PORT EVERGLADES ENVIRONS

PI: Dr. Cliff Bragdon and Dr. Carl Berkowitz

5.1 Summary

5.2 Introduction

5.3 Methods, Assumptions and Procedures

5.4 Results and Discussion

5.5 Conclusions

Appendix: Port Everglades Simulation

LIST OF FIGURES

Figures for Section 2.2

Figure 2.2.1: Communications Channels between Operator & RPUUV

Figure 2.2.2: Photograph of the RPUUV

Figure 2.2.3: The RPUUV Components

Figure 2.2.4: The RPUUV Component Layout

Figure 2.2.5: RPUUV Component Connectivity Flowchart

Figure 2.2.6: RPUUV Nose

Figure 2.2.7: Nose Section Components

Figure 2.2.8: The RPUUV Mid-Section

Figure 2.2.9: CAD Model of the Tail Section

Figure 2.2.10: Close-Up Photograph of RPUUV Tail Section

Figure 2.2.11: Exploded View of the Tail Section Components

Figure 2.2.12: RPUUV Top Level Wiring Diagram

Figure 2.2.13: Software Flow Diagram for the RPUUV Motherboard

Figure 2.2.14: State Diagram for the RPUUV Motherboard

Figure 2.2.15: RPUUV Tow Float Control Board Wiring Diagram

Figure 2.2.16: RPUUV Tow Float Software Flow Diagram

Figure 2.2.17: RPUUV Surface Tow Float

Figure 2.2.18: Initial Testing in a Shallow Pool

Figure 2.2.19: Screen Dump from Remote Desktop Application on Topside Laptop

Figures for Section 2.3

Figure 2.3.1. Overview of the RPUV control using a tow-float and electro-magnetic waves.

Figure 2.3.2. Detailed diagram of the RPUV control using a tow-float and electro-magnetic waves.

Figure 2.3.3. Functional diagram of the tow-float.

Figure 2.3.4. Overview of the RPUV control using a tow-float and acoustic waves.

Figure 2.3.5. Detailed diagram of the RPUV control using acoustic waves.

Figure 2.3.6. The remote control components of the RPUV.

Figure 2.3.7. High-level flow chart of the acoustic piloting and positioning at the user end.

Figure 2.3.8. High-level flow chart of the acoustic piloting and positioning at the RPUV end.

Figure 2.3.9. Acoustic remote piloting electronics (left) and positioning array (right).

Figure 2.3.10. High-speed high-frequency acoustic modem setup.

Figure 2.3.11. 43008-bit jpg image and 8192-bit pgm image transmitted during tests in the intra-coastal waterway.

Figure 2.3.12. Set of experiments performed at the FAU SeaTech Marina (Dania Beach, Florida).

Figure 2.3.13. Sky view of the south turning basin of Port Everglades and FAU SeaTech marina.

Figure 2.3.14. Sky view of the south turning basin of Port Everglades, indicating the source and receiver locations (25, 50 and 75 m ranges) and the scope of the source (10 m watch circle) during a series of experiments.

Figure 2.3.15. Boat view of the south turning basin of Port Everglades, indicating the source and receiver locations during the experiments of January 2006.

Figures for Section 2.4

Figure 2.4.1 – Depiction of Effect of Environmental Characteristics within the Port Environment

Figure 2.4.2 – Site Locations for EIS Taken by Environmental Protection Agency.

Figure 2.4.3 – Average Temperature Profile from CTD Stations at Port Everglades.

Figure 2.4.4 – Average Salinity Profiles from CTD stations at the Port Everglades.

Figure 2.4.5 – Box Plot of Turbidity Concentrations from Port Everglades.

Figure 2.4.6 – Chart of Port Everglades Representing the Four Regions.

Figure 2.4.7 – Turbidity Profiles at 4 representative locations within Port Everglades.

Figure 2.4.8 – Sound Speed Profiles at 4 representative locations within Port Everglades.

Figures for Section 2.5

Figure 2.5.1 Model of PZT disk transducer

Figure 2.5.2 Equivalent circuit of PZT transducer

Figure 2.5.3 One way beampattern of circular disk transceiver showing a one way -3dB beamwidth of 60 degrees.

Figure 2.5.4 An example of a sensor configuration for the collision avoidance

application and a flooded payload section.

Figure 2.5.5 Transceiver that generates acoustic signals and measures acoustic

returns for the collision avoidance sonar.

Figure 2.5.6 Encapsulated electronics package for 8 channel collision avoidance

sonar system.

Figure 2.5.7 Image of mine-like target resting on seabed.

Figures for Section 2.6

Figure 2.6.1 Wageningen Model 19A duct profile [3].

Figure 2.6.2 RPUUV Holder design and execution.

Figure 2.6.3 Tow carriage velocity profile.

Figure 2.6.4 Tow carriage main assembly and motor controller box.

Figure 2.6.5 Schematic of stepper motor controller and force transducer system.

Figure 2.6.6 Overall configuration of experimental setup.

Figure 2.6.7 RPUUV design.

Figure 2.6.8 RPUUV hydrodynamic model construction and assembly: tail section (top left); nose section (top right); partial assembly (bottom).

Figure 2.6.9 PIV system setup; hydrodynamic model mounted in test section.

Figure 2.6.10 Velocity field measurements at the aft section of the vehicle with propeller off and the vectored-thruster set to an angle of 0o tilt. Green arrows indicate direction and magnitude of velocity at several points in the flow.

Figures for Section 2.7

Figure 2.7.2-1. Illustrative sketch of the RPUUV

Figure 2.7.3-1. RPUUV motion: coordinate system and notations.

Figure 2.7.3-2:. 3DOF RPUUV motion in the horizontal plane

Figure 2.7.3-6-i. T = 2 [N], α=10o, xG=-0.1 [m], yG = 0 and without aft fins.

Figure 2.7.3-6-ii. T = 4 [N], α=50o, xG=-0.1 [m], yG = 0 and without aft fins.

Figure 2.7.3-7i. Simulated vehicle trajectories for Cases 1 and 2

Figure 2.7.3-7ii. Evolution of sway velocity v in time. Note that the spikes correspond to perturbation introduced at t = 50 [s]

Figure 2.7.3-7iii. Trajectories of the vehicle with and without aft fins for xG = +0.1 [m]

Figure 2.7.3-7iv. Time evolution of sway velocity perturbed at t = 50[s] with and without aft fins and with xG = +0.1

Figure 2.7.3-7v. Trajectories of the vehicle with and without aft fins for xG = -0.1 [m]

Figure 2.7.3-7vi. Time evolution of sway velocity perturbed at t = 50[s] with and without aft fins and with xG = -0.1 [m].

Figure 2.7.4-1. RPUUV motion in the vertical oxz plane.

Figure 2.7.4-4i. Motion in the vertical plane for T = 3 [N], β=10o, xG =+0.01 [m], zG =+0.03 [m] and t=(0,400 [s].

Figure 2.7.4-4ii. T = 5 [N], β=50o, xG =0 [m], zG =+0.03 [m] , and t=(0,300 [s])

Figure 2.7.4-5i. RPUUV motion in the vertical plane include tow-float cable force.

Figure 2.7.4-5ii Vehicle trajectory corresponding to T = 2.5 [N], β=0, xG=0, zG= 0.01 [m], τ =0.5 [N], δ = 45oand for values of xtow = -0.3, 0 and +0.3 [m].

Figure 2.7.4-5iii. Time evolution of pitch displacement (in radians) corresponding to T = 2.5 [N], β=0, xG=0, zG= 0.01 [m], τ =0.5 [N], δ = 45oand for values of xtow = -0.3, 0 and +0.3 [m].

Figure 2.7.4-5iv. Time evolution of pitch displacement (in radians) corresponding to

Thrust = 2.5 [N], β=0, xG=0.1 [m], zG= 0.01 [m], τ =0.5 [N], δ = 45oand for values of xtow = -0.3, 0 and +0.3 [m].

Figure 2.7.5-1. Modeling of the Aft Flow

Figure 2.7.5-2. Mapping of physical domain to uniform rectangular computational domain.

Figures Section 2.8

Figure 2.8.1 Example environments created in the simulator

Figure 2.8.2 The basic principle behind the ray tracing algorithms

Figure 2.8.3 The angle of reception of a reflected ray is measured from the axis of the transducer. This value is used to determine the directivity function that is applied to the signal.

Figure 2.8.4: A situation that would invoke the Emergency Reverse behavior

Figure 2.8.5: Disallowed Direction input membership functions.

Figure 2.8.6: Disallowed Direction output membership functions.

Figure 2.8.7: Flow chart for the Disallowed Direction fuzzy system.

Figure 2.8.8: Input membership functions for the Emergency Turn behavior.

Figure 2.8.9: Output membership functions for Emergency Turn

Figure 2.8.10: Input membership function for Goal Finding

Figure 2.8.11: Output membership function for Goal Finding

Figure 2.8.12: An example Disallowed Direction output

Figure 2.8.13: An example Goal Finding output

Figure 2.8.14: The combination of a desired direction and a not disallowed direction.

Figure 2.8.15: CLA defuzzification of an output membership function to arrive at a heading command for the vehicle

Figure 2.8.16: Surface describing the interaction of inputs and outputs in the Wall Following behavior.

Figure 2.8.17: Flow chart illustrating the interactions of behaviors in the OAS.

Figure 2.8.18: Transducer placement on the vehicle used for the results

Figure 2.8.19: Basic test results of the fuzzy OAS using a simple isolated cylinder.

Figure 2.8.20: Obstacle avoidance result using a short, flat wall.

Figure 2.8.21: Wall following behavior in a test tank.

Figure 2.8.22: The vehicle fails to escape the canyon if the Wall Following behavior is not activated.

Figure 2.8.23: The vehicle successfully reaches the goal point if the Wall Following behavior is activated.

Figures for Section 2.9

Figure 2.9.1 – WETStar Fluorometer.

Figure 2.9.2 – Eu/TTA peak excitation when scanning for 613 nm emission, with and without nitroglycerin.

Figure 2.9.3 – Eu/TTA emission in seawater, excited at 370 nm.

Figure 2.9.4 – Eu/TTA/OP emission in seawater, excited at 370 nm.

Figure 2.9.5 – Eu/OP/TTA emission in seawater, excited at 370 nm

Figure 2.9.6 – Response of all three europium compounds in seawater, with and without nitroglycerin, excited at 370 nm

Figure 2.9.7 – 613 nm fluorescence intensity as a function of solution methanol percentage.

Figure 2.9.8 – 613 nm fluorescence intensity as a function of solution methanol percentage.

Figures for Section 3

Figure 3.1 HDMAX camera

Figure 3.2a Pt. Hueneme image test target at 3 miles – long shot

Figure 3.2b Pt. Hueneme image test target at 3 miles – 5x zoom

Figure 3.2c Pt. Hueneme image test target at 3 miles – 10x zoom

Figure 3.3 Image of the HDMAX test at Port Everglades

Figure 3.4a ITC-developed solid-state store

Figure 3.4b SSR - open

Figures for Section 4

Figure 4.1. A generic video surveillance system: block diagram.

Figure 4.2. Block diagram of the proposed foreground segmentation system.

Figure 4.3. Movement related change detection.

Figure 4.4. "Stationary" change detection.

Figure 4.5. Change detection plot.

Figure 4.6. F-H algorithm: (a) the original frame, (b) the F-H output (k = 100) and without the post-processing step, and (c) the F-H output after post-processing that shows the undesired effect of losing the small boat.

Figure 4.7. Reducing the number of false positives with modified post-processing based on color moments and edge bins: (a) F-H segmented image without modified post-processing (79 segments), and (b) with our modified post-processing (15 segments).

Figure 4.8. Representative frame: (a) original frame, (b) reference background frame, (c) F-H segmentation, (d) final segmentation result.

Figure 4.9. Object Classification in a Visual Surveillance System.

Figure 4.10 Components and work flow of object classification.

Figure 4.11 Detecting ships by background subtraction and double thresholding with hysteresis.

Figure 4.12 Examples of Express Cruise.

Figure 4.13 Examples of Motor Yacht.

Figure 4.14 Examples of Recreational.

Figure 4.15 Examples of Speed Boat.

Figure 4.16 Examples of Sportfish.

Figure 4.17 Examples of Water Taxi

Figure 4.18 Demonstration of k Nearest Neighbor classification algorithm.

LIST OF TABLES

Table 2.6.1 RPUUV propulsion system design criteria

Table 2.6.2 Expected ducted propeller operating characteristics for straight-line motion:

Table 2.7.2-1 Principal Dimensions and Propulsion Estimates for the RPUUV

Table 2.7.3-1: Expressions for hydrodynamic forces and moments

Table 2.7.3-7i. Simulations 1 and 2

Table 2.7.3-7ii. Simulations 3 and 4

Table 2.7.3-7iii. Simulations 5 and 6

Table 2.7.3-7iv. Simulations 7 and 8

Table 2.7.3-7v. Simulations 9 and 10

Table 2.7.3-7vi. Simulations 11 and 12

Table 2.8.1: Parameters used to generate the maneuvers in the results

Table 4.1 Some of the Visual Surveillance Literature on Moving Object Detection

Table 4.2 Some of the Visual Surveillance Literature on Tracking

Table 4.3 Some of the Visual Surveillance Literature on Object Classification

Table 4.4 Some of the Visual Surveillance Literature on Object Behavior Analysis

Table 4.5 L1 Distance, Standard Voting, Leave One Out

Table 4.6 L1 Distance, Standard Voting, 10 Fold

Table 4.7 L1 Distance, Standard Voting, 10 Fold Stratified

Table 4.8 L1 Distance, Weighted Voting, Leave One Out

Table 4.9 L1 Distance, Weighted Voting, 10 Fold

Table 4.10 L1 Distance, Weighted Voting, 10 Fold Stratified

Table 4.11 L2 Distance, Weighted Voting, Leave One Out

Table 4.12 L2 Distance, Standard Voting, 10 Fold

Table 4.13 L2 Distance, Standard Voting, 10 Fold Stratified

Table 4.14 L2 Distance, Weighted Voting, Leave One Out

Table 4.15 L2 Distance, Weighted Voting, 10 Fold

Table 4.16 L2 Distance, Weighted Voting, 10 Fold Stratified

Table 4.17 Summary of Maximum Value of Accuracy in Each Experiment

Table 4.18 Artificial Neural Network, Ten-Fold Stratfied

Table 4.19 K Nearest Neighbour (k=4), L1 Distance, Standard Voting, 10 Fold Stratified

Table 4.20 Paired T-Test for Comparing k-NN and ANN

EXECUTIVE SUMMARY

The Center for Coastline Security Technology (CCST) focuses on research, simulation, and evaluation of coastal defense and marine domain awareness equipment, sensors and components. It builds upon the existing efforts and expertise in coastal systems and sensor research at the Institute for Ocean and Systems Engineering (IOSE), the Imaging Technology Center, the Department of Computer Science and Engineering and the University Consortium for Intermodal Transportation Safety and Security at Florida Atlantic University.

New technologies are needed to enhance surveillance and inspections of marine activities in the coastal zone that includes major ports, small inlets, beaches, remote coastal areas and their approaches. To be efficient and cost effective it is imperative to mount the surveillance systems and sensors on autonomous platforms that can operate unsupervised for extended periods of time. The task is to effectively integrate sensors with underwater, surface and airborne autonomous and remotely operated platforms and to incorporate video and image analysis and data mining methods to quickly and effectively identify threat events. These algorithms will automatically detect moving objects of interest, identify their location, track them through a video sequence, and classify them into semantic categories. In addition this group has unique resources and state-of-the-art capabilities in the area of real time 3D/4D virtual simulation that can be used to evaluate coastline security technologies and systems prior to deployment, and to carry out threat analysis using simulation for both training and exercises.

The three technologies that are being developed in this program are:

1) Underwater vehicles for survey and inspection: The faculty and staff in Florida Atlantic University’s Department of Ocean Engineering has extensive experience in developing, building and operating Autonomous Underwater Vehicles (AUVs). In the proposed program we will develop a low cost, one man operated, remotely piloted unmanned, untethered, underwater vehicle, which will provide real time underwater video and sonar images to a topside console. The specific application to be addressed is underwater inspections by rapid response teams, and routine inspection activities, currently carried out by scuba divers. This technology is intended to reduce the need for divers on a 24/7 basis.

2) High Definition Video Systems: High definition video cameras provide an order of magnitude improvement in field of view and/or range over and above conventional systems. Consequently they are a necessity for harbor surveillance, but their implementation in this environment is limited by size and cost. At Florida Atlantic Universities Imaging Technology Center, a compact high definition camera has been developed and is ready for the commercial market, the primary customers being the film industry. For the port security application there are several research issues that still need to be addressed, specifically, managing the high data output rate of the camera, and testing the camera in the marine environment. The test and evaluation issue has been addressed by the ITC center in collaboration with NAVSEA Carderock’s South Florida Test Facility, which has towers overlooking Port Everglades and the adjacent inlet that are already used by the USCG for video surveillance. The data management issue is being addressed by Florida Atlantic University’s Department of Computer Science and Engineering.

3) Simulation of Port Security Scenarios: For any port, the evaluation of threat scenarios, and the operational procedures required to implement appropriate countermeasures, cannot be fully enacted during training exercises. Further, the effectiveness of surveillance operations needs to be fully evaluated prior to installation of permanent equipment. This is best achieved through computer simulation of the port environment. Florida Atlantic Universities Center for Intermodal Transportation Safety and Security has recognized expertise and software available for this activity, and has initiated proof of concept studies at Port Everglades. In this program a simulation of the activities in Port Everglades has been developed, including the interaction between Navy assets and civilian activities.

This document is the final report for year one of this three year program and describes the progress on the following projects

• The Development of a Remotely Piloted, Unmanned, Untethered, Underwater Vehicle (RPUUV),

• High Definition High Frame-Rate Color Camera for Surveillance

• Video Analysis and Image and Video Data Mining

• Port Everglades Simulation

The project includes the activities of eleven principle investigators. The following provides a summary of the achievements of each element of the program.

Development of a Remotely Piloted Unmanned Underwater Vehicle

PI: Dr. Stewart Glegg, Project Manager: Robert Coulson

The development of the Remotely Piloted Unmanned Underwater Vehicle is described in Section 2.2. The objective of year one of this program was to develop a vehicle that communicated with a topside console through an RF link, and to test this vehicle in different environments with simple sensors such as an underwater video and a commercially available sonar system. The vehicle that has been developed features a vectored thruster with an 80 deg angular range, which allows the vehicle to maneuver in tight spaces. The weight of the vehicle is approximately 35 lbs and is easily launched and recovered by a single operator from the side of a small vessel. The vehicle includes an on board computer which processes the sonar data, the underwater video and the output from an on board compass, pitch and roll sensor. The data from these systems is relayed through a wireless RF link on the tow float to the topside console using a remote desktop capability, which is described in section 2.3. The vehicle is controlled through the RF link using a commercially available remote control device developed for model aircraft.

A complete description of the component layout in the vehicle, the thruster and control system, the onboard power supplies and sensor systems is described in this section. Also included is a description of the in water test which were carried out in May 2006 and recommendations for future development based on the results of those tests.

Acoustic and RF Communications

PI: Dr. P. Beaujean

The main objective of this portion of the project is to achieve communications for the purpose of transmitting and receiving information wirelessly between a user and the Remotely Piloted Underwater Vehicle (RPUV). Transmitted information is used to pilot the RPUV and relay its position. Information received from the RPUV combine acoustic images of the environment and status report of the vehicle. To achieve such goals, radio wave (WiFi) communication is achieved using a tow-float. Whenever the tow-float solution becomes impractical, a slower but fully wireless acoustic modem is to be used. The communication design must consider the issues associated with acoustic communications in port at high data rates, using a high-frequency acoustic modem, and the piloting and tracking of the RPUV, using a command-and-control acoustic modem.

Environmental Assessment and Modeling

PI: Dr. George V. Frisk

Daniel H. Sheahan and Dustin E. Whipple (Graduate Students)

The goal of this project is to characterize Port Everglades both acoustically and optically as these properties relate to the operation of the Remotely Piloted Unmanned Underwater Vehicle (RPUUV) technology being developed for the Center for Coastline Security Technology (CCST). Once the relation of these properties to the functionality of the RPUUV sonar and video systems is adequately understood, this approach can be applied to other port environments in which a similar surveillance system may be employed.

This report is one of the Year 1 Deliverables to the CCST project, and provides the results of a literature search on the current state of knowledge of the Port Everglades environment. Although there are many measurements of these properties in the surrounding areas offshore, there are none actually made within the Port itself. The archival information providing the best insight is an Environmental Impact Statement (EIS) prepared in 2004 that included both Port Everglades and Palm Beach Harbor. However all of these measurements were made offshore and are therefore primarily of interest as points of comparison with data obtained within the Port itself. Due to the lack of existing data, a sampling strategy was developed for acquiring new information about the Port, and an ambitious measurement program was initiated in Year 1 (one year ahead of schedule). This work was accomplished using the FAU research vessels to conduct 15 at-sea trips in which more than 175 profiles were measured using a Falmouth Scientific CTD (conductivity-temperature-depth) and Seapoint turbidity meter. This report presents a quick look at these data sets, with subsequent detailed analysis to take place in Years 2 and 3 of the CCST project.

Collision Avoidance Sonar,

PI: Dr. Steven Schock

A low cost collision avoidance sonar was designed to fit into small UUVs. The sonar provides the user with the ability to position transceivers along the nose and hull of the UUV and to configure the electronics package so it can be deployed with a wide variety of nose payloads. A compact transceiver with a circular piston was designed to mount flush to the skin of the UUV. The sonar has a processor that can be programmed to perform tasks such as notifying the UUV of obstacles in its path or the distance to a ship’s hull as the UUV conducts an underwater hull survey. A sea test in the Gulf of Mexico was conducted to verify the performance of the collision avoidance sonar electronics package.

Hydrodynamic design and integration of RPUUV hull-form and control surfaces

PI: Dr. K. D. von Ellenrieder

The objective of the research was to contribute to the design and efficient performance of the RPUUV developed at Florida Atlantic University for port and underwater ship-hull surveys. The report describes the selection of propeller and propeller duct design, design and fabrication of experimental models and model mounts, results, findings, and recommendations.

Given a required propeller thrust estimates of 2.5 [N] and an optimum motor operating condition of about 1500 [RPM]/8 [oz-in] torque, a suitable propeller and propeller duct design was selected (Wangeningen Model 19A duct profile with Roboesch 5.2 [inch] diameter, 4-bladed model propeller).

A hydrodynamic model of the RPUUV, as well as a model mounting system, which permits the model to be mounted in any pitch/yaw angle of up to ±60o, in increments of 1o, were designed and constructed. Particle image velocity experiments were conducted to study the velocity field at the stern of the vehicle. The experiments reveal that there is some slight flow separation upstream of the propeller inflow, which can likely be alleviated with simple modification of the aft section.

The findings and contributions of this research to the design and development of the RPUUV are discussed.

Hydrodynamics Analysis and Simulations for Design and Operation of a

Remotely-Piloted Unmanned Underwater Vehicle (RPUUV)

PI: Dr. P. Ananthakrishnan

The objective of the research was to contribute to the design and efficient performance of the RPUUV developed at Florida Atlantic University for port and underwater ship-hull surveys. The report describes formulations of the hydrodynamic and dynamic problem, solution methodology, specific problems considered, results, findings, and recommendations.

The vehicle drag at the design speed of 1 m/s was determined using empirical formulas including the ITTC formula for skin friction. Using Taylor wake fraction formulas, the thrust and power requirements of the propeller were determined. It was estimated that the vehicle drag at a forward speed of 1 [m/s] is 1.6 [N]. Required propeller thrust was estimated to be 2.5 [N] and delivered power to be 3.3 [W].

The open-loop dynamics of the vehicle was examined by analyzing the equations of motion obtained with respect to body-fixed coordinates. The hydrodynamic forces were determined in terms of hydrodynamic coefficients. The dynamic stability and maneuverability of the RPUUV while under horizontal and vertical planes of motion were analyzed for a range of parameters including forward speed, location of the center of gravity and various angles of vector thruster. The analysis confirmed the suitability of a vector thruster for the RPUUV and determined an optimum location on the RPUUV for the connection of the tow float cable.

A viscous-flow solver was also developed to study the stern flow of the vehicle in order to determine an optimum aft profile of the vehicle. The governing incompressible Navier-Stokes equations are solved using a boundary-fitted coordinates method. The operation of the propulsion was modeled using a novel pressure boundary condition. The pressure was determined based on the thrust developed by the propeller.

The findings and contributions of the research to the design and development of the RPUUV are discussed. The source codes developed under this project are included in the appendix section of the report.

RP UUV Navigation and Control

PI: Dr. Edgar An

Response time to a threat or incident for coastline security is an area needing improvement. Currently, the U.S. Coast Guard or local law enforcement is tasked with monitoring and responding to threats in coastal and port environments using boats, planes, and SCUBA divers. This can significantly hinder the response time due to the uncertainty in the threat or incident’s location. One solution to this problem is to use autonomous underwater vehicles (AUVs) to continuously monitor a port or to be launched in response to an incident or attack. The AUV must be able to navigate the environment without colliding into objects for it to operate effectively. Therefore, an obstacle avoidance system (OAS) is essential to the activity of the AUV. This report describes a systematic approach to characterize the OAS performance in terms of environments, obstacles, SONAR configuration and signal processing methods via modeling and simulation. A fuzzy logic based OAS is designed using a simulation created in Matlab. Subsequent testing of the OAS demonstrates its effectiveness in unknown environments.

Chemical Sensors

PI: Dr. Richard Granata

This section describes the formulation of a chemical method to detect underwater trace explosives, as well as the design of a field-deployable device to implement the chemical method. The research goals are identified, the primary test materials, equipment and experiments are described and the results are discussed. The chemical compound, europium thenoyltrifluoroacetone, has been identified as an integral part of a viable underwater chemical detection method for underwater explosive traces.

High Definition High Frame-Rate Color Camera for Surveillance ,

PI: Dr. William E. Glenn

The objective of this segment of the project was to develop a high definition, high frame rate color camera for surveillance. A 3840x2160 30P color CMOS camera with variable frame rate and remotely controlled infrared filter changers was designed, fabricated, tested and demonstrated. The camera gathers 50 times the amount of information in its field of view as with standard resolution video cameras. A memory unit, solid-state store, for large amounts of imaging data generated by the camera was also designed, fabricated and tested. Field tests demonstrated that the camera’s high resolution makes it possible to do electronic zoom on sections of an image without the excessive loss of resolution present with standard cameras, and the high frame rate allows the use of moving target indication, velocity measurement and the observation of brief events that help classify suspected targets. It is recommended that work continue to upgrade the camera system and to add compression to the solid-state memory to allow increased record time. The HDMAX camera and solid-state recorder are ideally suited for video surveillance on ships, submarines, harbors, AUVs and drone aircraft.

Video Analysis and Image and Video Data Mining

PI: Dr. Borko Furht

This report summarizes the first year of research activities in the field of image and video analysis algorithms for coastline security. Our work has been focused primarily on algorithms and techniques for motion detection, object tracking, and object classification in maritime scenes. After having investigated existing algorithms in the literature, we proposed and implemented robust algorithms for scene segmentation, object detection, tracking, and classification in video sequences with complex, moving background.

The goal of a visual surveillance system is to detect abnormal object behaviors and to raise alarms when such behaviors are detected. After moving objects are detected, it is essential to classify them into pre-defined categories, so that their motion behaviors can be appropriately interpreted in the context of their identities and their interactions with the environment. Consequently, object classification is a vital component in a complete visual surveillance system.

In this project on coastline security, the objects of interest are ships. We formulate the task of ship classification in the standard framework of pattern recognition:

1) 402 instances of ship regions were collected from surveillance video, and were classified into 6 types by human observers.

2) The shape feature of each region was extracted using MPEG-7 region-based shape descriptor.

3) A classification algorithm is applied to classify ships based on the similarity of their shape features.

4) The classification performance was evaluated using cross validation and the optimal parameters were determined.

We have applied two commonly used classification algorithms (k-Nearest-Neighbor algorithm and artificial neural network) and compared their performance based on the mean accuracy of ten stratified ten-fold cross validation.

For k-Nearest-Neighbor algorithm, we have performed combinatorial experiments to examine the effect of the following factors on classification performance: version of k-Nearest-Neighbor algorithm (standard voting or distance-weighted voting), type of distance measure (L1 or L2), and type of cross validation (leave-one-out or ten-fold or stratified ten-fold). The experimental results do not reveal any significant performance differences between standard voting and distance weighted voting. L1 distance (city block distance) is preferred to L2 distance (Euclidean distance) for computing the similarity of shape between ship regions, and the recommended number of nearest neighbors is 4. The recommended parameters for the shape descriptor are: m = 24 (the number of angular directions); n = 12 (the number of radial scales). The classification accuracy of k-Nearest-Neighbor algorithm based on stratified ten-fold cross validation is about 91%, which compares favorably with existing work. k-Nearest-Neighbor algorithm outperforms artificial neural network, and takes less computation time.

The proposed classification procedure based on MPEG-7 region-based shape descriptor and k

Nearest Neighbor algorithm has the following advantages:

• MPEG-7 region-based shape descriptor is robust to noise and tolerant to objects with holes and objects fragmented into several components. It can be applied to extract shape features from not only ships, but also other rigid objects, such as airplanes, vehicles, etc.

• k-Nearest-Neighbor algorithm is scalable to the size of the data set. When new labeled cases are added to the case library, it is not necessary to re-build the classifier, which is often computationally expensive for other classification algorithms.

• k-Nearest-Neighbor algorithm can provide a meaningful interpretation of the classification results by showing the similar cases retrieved from the library.

• The computation cost of k-Nearest-Neighbor algorithm is reasonable in comparison to other algorithms.

The proposed classification procedure can be trained off-line to tune the parameters for a specific surveillance domain, and then integrated into a real-world visual surveillance system as the object classification module. In a production system, the input to our module should be a binary image highlighting the region corresponding to the object to be classified. The input will be provided by the object detection and tracking modules, which are currently being developed by other members of our team. The output is a classification label corresponding to a predefined object type. Since we have not used any constraints specific to ships, the module can also be applied to the classification task of other rigid objects, such as vehicles, airplanes, etc.

Virtual Port Simulation-Port Everglades Environs

PI: Dr. Carl Berkowitz and Dr. Cliff Bragdon

Virtual simulation is able to replicate the entire infrastructure and transportation activity (moving people and goods) through a photo-realistic computer digitally based real-time presentation of both existing conditions and alternative future conditions in a port. Through this simulation process, the port will be able to identify vulnerabilities, create realistic scenes for examination, view elevations at various multi eye-points, and evaluate advanced technologies. This technology, also, allows for the development of surface and underwater scenes in order to evaluate incident response training, conduct tabletop exercises, train management personnel and more effectively develop transportation security systems.

This report discusses the methodology used to build the Port Everglades virtual simulation model, including: data acquisition, using GIS information, inputting CAD files, overlaying intermodal traffic, converting information into a digitized data base, volumetric development and massing of infrastructure, texturing of fixed and mobile objects, visualization of climatic conditions, depicting time of day, presenting seasons and sun/moon angles, and development of runtime model (motion-based) in real-time.

1.0 INTRODUCTION

1.1 Overview

1.1.1 Background

The Center for Coastline Security Technology (CCST) focuses on research, simulation, and evaluation of coastal defense and marine domain awareness equipment, sensors and components. It builds upon the existing efforts and expertise in coastal systems and sensor research at the Institute for Ocean and Systems Engineering (IOSE), the Imaging Technology Center, the Department of Computer Science and Engineering and the University Consortium for Intermodal Transportation Safety and Security at Florida Atlantic University.

New technologies are needed to enhance surveillance and inspections of marine activities in the coastal zone that includes major ports, small inlets, beaches, remote coastal areas and their approaches. To be efficient and cost effective it is imperative to mount the surveillance systems and sensors on autonomous platforms that can operate unsupervised for extended periods of time. The task is to effectively integrate sensors with underwater, surface and airborne autonomous and remotely operated platforms and to incorporate video and image analysis and data mining methods to quickly and effectively identify threat events. These algorithms will automatically detect moving objects of interest, identify their location, track them through a video sequence, and classify them into semantic categories. In addition this group has unique resources and state-of-the-art capabilities in the area of real time 3D/4D virtual simulation that can be used to evaluate coastline security technologies and systems prior to deployment, and to carry out threat analysis using simulation for both training and exercises.

This effort includes activities at Florida Atlantic University's SeaTech campus, allowing researchers to leverage the existing U.S. Navy marine test & evaluation facilities, geographically combined with the adjacent major seaport at Port Everglades. This provides a unique land and aquatic test bed. Initial studies have focussed on acoustic sensors and high definition underwater and surface video mounted on unmanned fixed or mobile platforms. Emphasis has also be given to the development of optimal platforms for the efficient collection and integration of the information from multiple sensors.

1.1.2 Technical Objectives

As time progresses it is becoming increasingly apparent that providing elevated homeland security in ports and harbors is limited by operational costs. Budgets for port security are several times larger than they were before the events of 9/11/01 and cost is now a major issue for both federal and local agencies. Furthermore, when Navy ships dock in areas also used for civilian activities, security issues are more complex and require close collaboration between all agencies involved. The same principles apply in overseas ports, as evidenced by the attack on the USS Cole, and port security technology, which is portable to international locations, has an important role in force protection.

Given these prerequisites it is the primary objective of this program to develop new technology for port security that provides unique capabilities for security inspections, threat detection and rapid response, at lower operational costs. To achieve this, attention will be focused on three technologies in which the members of the center have existing expertise, with the intent of turning these technologies into operational systems in a three year program.

The three technologies that are being developed in this program are:

1) Underwater vehicles for survey and inspection: The faculty and staff in Florida Atlantic University’s Department of Ocean Engineering has extensive experience in developing, building and operating Autonomous Underwater Vehicles (AUVs). In the proposed program we will develop a low cost, one man operated, remotely piloted unmanned, untethered, underwater vehicle, which will provide real time underwater video and sonar images to a topside console. The specific application to be addressed is underwater inspections by rapid response teams, and routine inspection activities, currently carried out by scuba divers. This technology is intended to reduce the need for divers on a 24/7 basis.

2) High Definition Video Systems: High definition video cameras provide an order of magnitude improvement in field of view and/or range over and above conventional systems. Consequently they are a necessity for harbor surveillance, but their implementation in this environment is limited by size and cost. At Florida Atlantic Universities Imaging Technology Center, a compact high definition camera has been developed and is ready for the commercial market, the primary customers being the film industry. For the port security application there are several research issues that still need to be addressed, specifically, managing the high data output rate of the camera, and testing the camera in the marine environment. The test and evaluation issue has been addressed by the ITC center in collaboration with NAVSEA Carderock’s South Florida Test Facility, which has towers overlooking Port Everglades and the adjacent inlet that are already used by the USCG for video surveillance. The data management issue is being addressed by Florida Atlantic University’s Department of Computer Science and Engineering.

3) Simulation of Port Security Scenarios: For any port, the evaluation of threat scenarios, and the operational procedures required to implement appropriate countermeasures, cannot be fully enacted during training exercises. Further, the effectiveness of surveillance operations needs to be fully evaluated prior to installation of permanent equipment. This is best achieved through computer simulation of the port environment. Florida Atlantic Universities Center for Intermodal Transportation Safety and Security has recognized expertise and software available for this activity, and has initiated proof of concept studies at Port Everglades. In this program a simulation of the activities in Port Everglades has been developed, including the interaction between Navy assets and civilian activities.

In the following sections the details of each part of this program will be outlined and the following projects are described

• The Remotely Piloted, Unmanned, Untethered, Underwater Vehicle (RPUUV),

• High Definition High Frame-Rate Color Camera for Surveillance

• Video Analysis and Image and Video Data Mining

• Port Everglades Simulation

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download