Home | Contact ST  

Feature Article

USV-Based Security System For Civilian Harbors
The Swarm Management Unit Combines A Real-Time Motion Planner With Preplanned Positioning Optimization

By Enrico Simetti
Ph.D. Candidate
Dr. Alessio Turetta
Researcher
and
Giuseppe Casalino
Professor Department of Communication,
Computer and System Science
University of Genova
Genova, Italy



Safeguarding civilian harbors against terrorist attacks from the so-called “blue border” (i.e., the sea side of the harbor) is receiving increasing interest, especially after September 11. Multiple governments and international institutions have expressed the need for innovative and reliable technologies that could improve current harbor security infrastructures. NATO’s Defense Against Terrorism program has indicated harbor security as one of its 10 main areas of work, and this is an area where advanced technologies such as efficient sensor networks, electro-optical detectors and unmanned vehicles can prove to be very helpful.

In this context, the use of a team of unmanned surface vehicles (USVs) represents a promising solution for reducing harbor vulnerability. Under normal conditions, USVs can perform patrolling surveys of the more crucial waterways, providing a remote operator with acoustic and/or optical images acquired through onboard sensory equipment. Whenever a possible menace (i.e., an unauthorized vessel or a vessel moving in a suspect way) is detected, a USV can intercept and classify the vessel without exposing humans to a direct threat.

Previous experiments on USVs were mainly performed in the open sea or in closed waterways, usually in the absence of unknown moving vessels. Civilian harbors are instead dynamic scenarios where ship traffic may be intense and not totally known beforehand. Additionally, USV maneuvers must not perturb the operations of tourist or merchant ships; collision with other vessels risks personal injury and property damage.

In order to solve these problems, the Department of Communication, Computer and System Science (DIST) at the University of Genova and SELEX Sistemi Integrati (Genova, Italy) are jointly working to develop the swarm management unit (SMU), software that could supervise the operations of a team of USVs performing semiautonomous surveillance activities within civilian harbors.

The SMU, which is integrated into a command and control center developed by SELEX, provides decision support to supervisors by suggesting the optimal USV and the optimal path to intercept a suspect vessel; final decisions, however, are always made by human operators. The ongoing project, which began in 2008, is in the process of building a fleet of small, low-cost prototype USVs for testing SMU algorithms, which will be followed by testing with full-size USVs.


Asset Protection Scenario
In the considered scenario, an asset is a site of particular interest within the harbor area that must be protected against possible attacks from the sea. In order to protect the asset, every USV performs surveys by autonomously following preplanned paths selected by the SMU that optimize area coverage. The USV transmits images taken by onboard sensors to an operator working in a command and control center.

Whenever the operator detects some presumed anomalies, he or she can assume direct control of one or more USVs and drive them toward the point of interest for a deeper analysis of the suspect situation.

A different, more critical situation arises whenever a menace is detected by the harbor radar system. In such a case, one USV has to intercept the menace in the shortest possible time. In order to maximize the chances of success, two problems have to be tackled and solved.

First, once a menace is detected by the harbor radar, the best USV for interception has to be selected on the basis of the contingent situation (current positions, speeds and headings of the USVs, the menace and other vessels in the area).

Second, the nominal positioning of the USVs needs to be preplanned in order to increase the possibility of intercepting the menace before it passes the “security distance,” a distance close enough to the asset that an adequate response cannot be guaranteed.


Online Interceptor USV Selection
In the ideal scenario, where only the menace and the USVs are present, the problem could be solved by simple kinematic considerations based on position and motion of the menace and on the positions and speeds of the available USVs.

The harbor scenario is considerably more complex, as other vessels in the area represent fixed or moving obstacles that have to be avoided by the interceptor USV. This fundamental problem was addressed in a previous phase of the SMU project, completed in 2009 after about a year of work.

To solve it, an efficient real-time motion planner has been implemented that accounts for vessels in the area. The planner a priori knows the characteristics of the available USVs (maximum speed, minimum steering angle, sensory equipment) and receives the information extracted from the radar tracks every second (i.e., position, speed and heading of all vessels in the harbor area). The planner then computes the predicted time of interception of the menace together with the corresponding time-optimal path to reach it in a reliable way. The SMU suggests to the operator the most suitable USV for the mission. Once the operator selects a USV (be it the one suggested by the SMU or another one), the SMU immediately sends a list of waypoints for the selected USV follow.

When the menace is intercepted, if it is classified as hostile, it will trigger the harbor emergency procedures; otherwise nothing happens and the USV returns to its usual patrolling activity.

As the environment can dynamically change while a USV is following a reference path (because some of the other vessels might change their speed or heading), the motion planning procedure generates a new corrected time-optimal path anytime a variation of a radar track is detected.


Offline Positioning Optimization
The most recent phase of the project to be completed is the offline problem of deciding the optimal size and placement of the USV fleet. This problem requires some preliminary considerations to be drawn.

First, the statement that a menace can appear for the first time anywhere in the considered area makes the problem of asset protection intractable, since the menace could theoretically appear when it is already on the asset, making the USVs useless. It is therefore assumed that any menace is detected for the first time before it reaches a certain distance r (greater than the security distance) from the asset. Such an assumption is not unreasonable, as it can be assumed that an adequately rich suite of sensors can be built that would detect the menace before reaching this point.

Furthermore, although actual interception time of a real menace is dependent on vessel traffic, when considering offline optimization, no other vessels are assumed to be in the area; their presence is highly time-variant and cannot be predicted in a meaningful way.

Finally, as the velocity and the direction of any possible menace cannot be predicted, it is assumed that the menace is always moving with a given upper-bounded speed, heading toward the asset while avoiding static obstacles.

Even under these assumptions, the problem of deciding the “best” USV positioning is not straightforward. Indeed such a problem can be considered as a special instance of the so-called spatial resource allocation problem, which has been studied for many years and has registered interesting results, in particular in the field of fixed or mobile sensing networks.

However, unlike these cases, in the given harbor security scenario, it is not easy to identify a unique optimal criterion driving the optimization process. Several reasonable alternative strategies can be adopted in response to the specific index of performances one wishes to optimize.

The proposed solution is based on the following two priority-based criteria. First, the team of USVs must be able to intercept any menace before it can reach the security distance. Such a statement can be translated into a worst-case scenario optimization problem. It has to be granted that, even if the menace is detected in the closest possible position to the asset (i.e., at distance r), at least one USV is always in the position to intercept the menace before it reaches the security distance.

In order to solve such an optimization problem, a series of Monte Carlo simulations are performed. For a given distribution of USVs in the considered area, several stages of the Monte Carlo simulation are executed by varying the initial position of the menace. At each stage the set of interception points related to each USV is calculated, and the one at the greater distance from the asset is selected. Then, by changing the menace’s position, the worst-case scenario is found. If the resulting worst-case distance is acceptable (i.e., if it is greater than the security distance) the simulation is concluded and the current USV distribution meets the safety requirements. Otherwise the position of the fleet of USVs is modified through a gradient descent technique and the process is repeated for a given number of steps.

If the worst-case interception distance is still lower than the minimum required one, the given number of USVs is not sufficient to always guarantee reaching the menace in time. The problem therefore has to be solved by increasing the number of USVs until the required threshold is met. The resulting number, k, of USVs is then the minimum number required.

In case the number of available vehicles N is greater than k, the remaining USVs can be exploited to solve a secondary optimization problem: minimizing the maximum interception time. It is easy to see that this second criterion allows the extra vehicles to spread out in an area to safeguard the asset, making the team more reactive against menaces detected at any distance from the asset greater than r.


Conclusions and Future Work
SMU is a tool for controlling the operations of a team of USVs performing surveillance activities within civilian harbors. In order to protect an asset and intercept menaces in a crowded harbor environment, the SMU works in two parts, the first running online and the second computed offline. The first component involves selecting the time-optimal USV for intercepting a detected menace before it comes too close to the asset. The second allows the SMU to optimize all of the USVs’ nominal positioning to guarantee an adequate level of security at all times.

Ongoing work involves tackling the problem of multiple assets and/or multiple menaces. Another functionality that can and should be added is the implementation the Convention on the International Regulations for Preventing Collisions’ navigation rules for the USVs.

Even more important are field tests. Engineers are in the process of building a fleet of low-cost USVs that are about one to 1.5 meters long, equipped with only basic sensors to provide navigation functionalities.

The latest version of the SMU will go through sea trials in the La Spezia, Italy, harbor in spring 2011. After these tests, full-size USVs (between five and seven meters long) with a full sensor load will be built and tested.


Acknowledgments
The authors would like to thank Enrico Storti and Matteo Cresta from SELEX for their fundamental contributions.



Enrico Simetti, a Ph.D. candidate at the Department of Communication, Computer and System Science at the University of Genova, is a computer science engineer. His current research work is focusing on automated security systems in the harbor field, cooperative multirobot systems and embedded systems.

Dr. Alessio Turetta is a researcher at the Department of Communication, Computer and System Science at the University of Genova, focusing on control systems and embedded systems, and a mechatronic specialist with Graal Tech S.r.l. He has worked mainly in underwater robotics.

Giuseppe Casalino is a full professor at the Department of Communication, Computer and System Science at the University of Genova. His research activities are mainly in the field of robotics and automation, with a special focus on planning, motion and interaction control problems within sensorized multirobot structures.




-back to top-

-back to to Features Index-

Sea Technology is read worldwide in more than 110 countries by management, engineers, scientists and technical personnel working in industry, government and educational research institutions. Readers are involved with oceanographic research, fisheries management, offshore oil and gas exploration and production, undersea defense including antisubmarine warfare, ocean mining and commercial diving.