Networked Robotics System
In the initial testing phase of any system, simulations offer an edge over real experimentation. They are low-cost, visual, fast, scalable and can be reconfigured to suit new experiments. Simulations play an important role in academia and industry as a necessary research tool to analyze and justify proposed theoretical models. There has not been any remarkable attempt to simulate mobile agents for networked robots. Although Webots© [154] and CARMEN [155] are powerful tools for simulation of multi-robot system applications, mobile agent middleware for multi-robot or networked applications such as described in [73] and [74] have not been transferred to the simulation environment.
In order to study the performance of agent migration and cloning mechanisms developing a simulator for simulating mobile agent based networked robotic scenarios thus became imperative. This chapter reports the efforts taken in developing a simulator that allows mobile agents to move around a network of nodes, service them and clone on demand. The nodes can generate requests for service and hence are synonymous to robots. They can also host mobile agents and provide for all related computations thus simulating the quintessential middleware. The simulator thus supports the architecture depicted in Figure 2-3 of Chapter 2. Simulations carried out could thus be viewed to be those for a mobile agent based networked robotic system. The simulator provides the number of step-counts taken to reach or achieve the desired goal, the number of hops taken as also the intra-node computations and inter-node
communications – all of which can be used to judge the performance of the mechanisms being simulated. The performance measurements of the existing and proposed mobile agent related mechanisms described in earlier chapters of this thesis were found using this simulator.
6.1 Salient Features of the Simulator
The simulator developed is basically a Java application that can simulate mobile agents and mobile robotic nodes. It can simulate all the seven migration mechanisms (viz. Random, Conscientious, EVAP, CLInG, G-B, PherCon and PherCon-C) and also the cloning controller described in earlier chapters. Agent behaviours and its mobility along with robotic node movement relative to one another can be simulated through this application. The salient features of this simulator are described below.
1. Robotic Nodes: Each node in the simulated network performs all actions pertaining to a robot. These include requesting for services (acting as an RRS), pheromone management and discovering its one-hop neighbours in dynamic environments. The robotic node also doubles to act as the middleware (explained in Section 2.4 of Chapter 2) and hosts the mobile agents and performs the relevant agent related tasks.
2. Mobile Agent Support: The simulator allows for mobile agent to migrate and clone. These agents and clones may also be removed based on specified conditions.
3. Inter-entity communications: The simulator supports inter-entity communications that include – Agent-Agent, Agent-Robotic node and Robotic node-Robotic node interactions.
4. Network Topology: Simulations can be carried out for custom network topologies. The simulator need only be given the coordinates of the robotic nodes through an input file. This facility comes handy when one needs to test a
variety of topologies and situations. Other inputs specifying the exact step-count when a robotic node should become an RRS can also be provided. Each node has a communication radius which is used to decide which nodes can form the links and communicate. Two nodes are linked only if they fall within each other’s communication radius.
The simulation environment supports both static and dynamic scenarios.
The topology of the network in a static scenario is fixed and is based on the initial positions of the robotic nodes. In dynamic scenarios, the underlying topology can be made to vary in a random manner during the simulation run.
The nodes act as mobile robots thus simulating a Mobile Ad hoc Network of Robots (MANER) [112]. Their interconnecting links with other nodes are formed and broken as they move.
5. Logging: All results available on the simulation console as also the intermediate states and data pertaining to the entities of the network can be recorded and saved onto log files for off-line analysis.
6. Rendering: A visual presentation of the simulation is provided by the simulator. The visualization helps in quickly observing the behaviour and migration of the mobile agents and robotic nodes. In single-stepping mode the simulator allows the user to trace the manner in which both the agents and the nodes perform. A screenshot of the rendered simulation window is shown in Figure 6-1 has a total of 200 nodes. The white squares within denote robotic nodes that can host the mobile agents while the red squares denote the RRSs.
The yellow spots on the white squares denote the mobile agents which migrate to their neighbours using one of the chosen migration strategies. Faint circles around each node denote the wireless communication range of that node. The green segments between the nodes indicate that they can communicate with each other. The yellow colored links indicate the pheromones diffused across he neighbours. In a dynamic scenario the nodes can be seen to move in random directions.
Fig. 6-1 Screenshot of the window rendered by the simulator
The GO button starts the simulation and stops only when the termination condition, such as servicing a pre-determined number of RRSs, is reached. The STEP button allows for stepwise execution. The buttons designated NEIGHBOUR, AGENTINFO and PHEROMONES facilitate the storing of the pertinent information in specific files. The simulator also offers status
information such as the current migration strategy of the mobile agents, Number of Nodes and Edges, and the Mode of simulation.
6.2 System Requirements
The simulator can be run on either Windows or Linux based computing systems provided Java Development Kit (JDK) version 1.7.x and JAVA 2D Graphics visualization tool are installed.
6.3 Conclusions
This chapter provided a brief description on the simulator developed for the purpose of testing the proposed bio-inspired mechanisms. It also highlighted the salient features of the simulator. It was observed that the simulator could be used to test algorithms used in many other domains of networks. This simulator could prove to be a useful tool in the study of algorithms related to a myriad of areas, with minimal modifications. These areas include –Resource discovery in Ad-hoc networks, Sensor network protocols and Network management using mobile agents. Though simulations play a vital role in proving the efficacy of algorithms and mechanisms, the underlying assumptions that form the basis of simulation can make it portray results that need not hold good in the real world.
This is especially true in the realm of robotics where closed world simulated solutions cannot be directly thought of to be true. It was thus necessary to carry out real world implementations of the proposed algorithm to prove its efficacy and viability. The succeeding chapters portray the attempts made towards the realization of a physical working model of a robotic network that uses the proposed PherCon-C mechanism. It also describes how the very same mechanism could provide and support the existence of an Internet of Things.