SURV-TRACK

In this chapter, we first propose a model for the multi-robot indoor surveillance application supported by a wireless sensor network. Then, we tackle two different underlying issues in the SURV-TRACK application and propose solutions with respect to the following features (1) multi robot task allocation for target capturing and (2) path planning. 

This section presents an overview of the system model of the SURV-TRACK application and the underlying challenges to be addressed. 

Figure 1 illustrates the model of SURV-TRACK application, which consists ofthree main components ... 

The interaction between these three components depends on the coordination strategy. However, in our model, the WSN should be active in the sense that it contributes to the intelligence of the entire system. Unlike the case of autonomous robotic systems, where intelligence is embedded in the robots, one particular goal of SURV-TRACK is to migrate the intelligence—or part of it— to the WSN to help make the correct decision and build efficient plans. This issue will be presented in detail in Section 4 and Section 5. 

The design of the SURV-TRACK application raises several challenges. There aret5 ically three underlying questions that should be answered in the SURV-TRACK framework (1) How should the system components effectively communicate with each other (2) How can the tasks be efficiently subdivided among the different actors (mobile robots and sensor nodes) of the system (3) How can a mobile robot plan its path to accomplish its mission ... 

The network architecture ofthe SURV-TRACK model plays a central role in the performance of the system. In fact, the surveillance application encompasses cmcial requirements that must be carefully considered in the design of the network architecture ... 

In what follows, we describe the main ehar-acteristics of the network model of the SURV-TRACK application and we discuss the potential standardized technologies that fit the application. To meetthe aforementioned requirements, we propose atwo-tier architecture forthe SURV-TRACK application, as depicted in Figure 2. 

A Service-Oriented Approach In the SURV-TRACK application, mobile robots are responsible for providing target tracking and capturing. However, each mobile robot makes part of ateam, and the task assigned to a robot in a collaborative mission depends on the mission nature and on the tasks assigned to the other robots teammates. It... 

There are roughly two possible approaches to implementing SOA in SURV-TRACK (1) the Web Services approach, (2) the Representational State Transfer (REST) approach (Fielding, 2000). In what follows, we give a brief overview ofthese two approaches and discuss their advantages and drawbacks. 

Web Services technology is the leading implementation of SOA on the Web and can be used to specify service description, publication, and discovery. In the literature, several research papers propose using Web Services in robotic applications (Kim,etal., 2005 Levine Vickers, 2001). In the context of SURV-TRACK, amobile robot defines... 

The REST-based approach is another paradigm of SOA implementation, known as Resource Orientation Architecture (ROA), which follows the classical client/server model. It considers a service as a resource that can be accessed through the Web using traditional HTTP requests. These resources can be manipulated using the CRUD (Create, Read, Update, Delete) style via the typical HTTP methods, namely GET to request/retrieve a resource, POST to modify/update a resource, PUT to create a resource and DELETE to remove a resource. In the context of SURV-TRACK, the service provider or simply the server i.e. amobile robot or a sensor node) creates a resource for each service they provide and makes it accessible to service consumers or simply clients (i.e. control station) through a Uniform Resource Identifier... 

In Section 3, we presented the main networking and application layer protocols and system requirements of the SURV-TRACK application. These provide useful guidelines for any real-world implementation of an indoor surveillance application encompassing mobile robots and a sensor network. The aforementioned netwoik and application layers architecture offers ameanstothe different components of the surveillance system to communicate however, it is necessary to devise intelligent coordination mechanisms between mobile robots, sensor nodes, and the control station to effectively accomplish the taiget capturing missions. This is the purpose of the next section. 

It just happens that many multi-robot task allocation problems are NP-Hard problems these problems have no exact solutions in a polynomial time, therefore heuristics and approximate algo-rilhms are typically used to estimate solutions. The purpose of tiiis chapter is to propose solutions to this problem and apply them to the SURV-TRACK application. 

In the following section, we present solutions to the MRTA problem for SURV-TRACK based on the three aforementioned approaches. 

We are currently working towards implementing a real-world prototype of the SURV-TRACK application using a team of Wifibot Lab robots supported by a TelosB sensor network. This will help us assess the real behavior of the coordination between robots and sensor nodes and provide an experimental framework to design, implement and validate solutions to both MRTA and path planning problems. 

---