Introduction
    People prefer to send robots instead of humans to fulfill tasks in hostile and/or hazardous environment such as Nuclear or Biochemical contaminated areas, Battlefield, or Deep under water. The robots should be able to help humans by monitoring, exploring the environment, cooperating with humans or with other robots in order to cover the largest area in a short time and complete the task as soon as possible. 

    Two kinds of robots are proposed for large-scale hostile area: individual expensive, complex robot (Figure 1) and large number of cheap, simple mobile robots (Figure 2). Many studies [2,3,4,5] indicated that employing multiple inexpensive, simple mobile robots as opposed to a single expensive, complex mobile robot could perform tasks in a large-scale area more efficiently. By dividing the task into many sub-tasks and synchronously working on the sub-tasks, the robots can complete the task faster than a single robot. Multiple robots also introduce redundancy for better fault tolerance than single robot [2,3]. However, the absence of collaboration in the robot team introduces the risk of interference between robots. This interference can reduce the performance of robots teams. To make multiple robots collaboratively execute complicated and sophisticated tasks, a collaborative control-architecture needs to be developed [3].

Goal of CMR Project
    The CMR project is mainly focused on the coordination and control of large-scale distributed robotic systems. The goal of the project is to use multiple small, low-cost robots, with a limited range of local communication ability, to collaboratively search and engage tasks in an unknown large-scale hostile area. Figure 3 is an overall research framework for the CMR project. Currently, are we working on three focused areas: The multiple robotic collaborative algorithm research, Robotic navigation control algorithm research and Real ER1 robots collaboration implementation. 

The multiple robotic collaborative algorithm research

    The problem being addressed in this research is how to cooperatively search and engage an indeterminate number of tasks in a large-scale hostile area using multiple cooperating mobile robots with limited range communication. To solve the problem, we have developed a collaborative approach for dynamically, automatically and quickly formatting small, lightweight, low-cost robots into functional teams to collaboratively search and engage tasks in a large-scale hostile area without central control and global communication [4]. The key concept of the approach is swarm behavior. By applying the three properties of the swarm behavior: separation, cohesion and alignment, our approach can ensure the robot group attains large region coverage, dynamically stable connectivity and fast targets approach [5].

　

    A multi-agent platform, Madkit, is used for implement the algorithm. Madkit is a software framework fully implemented in the Java language. It simplifies the implementation of multi-agent systems through a middle-ware that claims to comply with the FIPA specifications. The only system requirement is the Java Run Time version 1.2 or above. Two research applications are implemented on this simulation platform:

1. Dynamically detecting the hazardous materials contaminated areas’ perimeters by multiple robotics. (Figure 5)

2. A Multi-agent Cooperative Strategy for Search and Rescue in The Large-scale Disaster  Environment. (Figure 6)

　

　

Robotic navigation control algorithm research

      One Robotic simulator (Play-stage) is used to demonstrate our research on the robotic navigation control algorithm. Player/Stage refers to two software packages produced by the Robotics Research Laboratory at the University of Southern California. Using Stage has several advantages for our purposes.  First, it has a GUI interface so that the simulation can be viewed graphically – which means that a simulation engine does not need to be developed for this project from scratch. Second, it allows for ideas involving more and different robots and sensors than are available physically.  Finally, it protects the robots from the inevitable implementation problems (it does not hurt a Stage Player to drive into a wall or off a cliff, for instance).

　

    Our research in this session is finding a collaborative control algorithm to control one or more robots equipped with different kinds of sensors, such as sonar, IR sensors and video camera exploring an area with unknown number of obstacles [7]. Reactive behavior and Fuzzy Logic algorithms are used for helping robots avoid obstacles and find the targets.

　

       

Implement the collaborative algorithms on the real robots.

    To verify the efficiency of the algorithms in previous researches, the collaborative algorithms that succeed in the simulation will be implemented on three real ER1 robots [8] in our lab. The robots are equipped with ultrasonic scanners and GPS for distance measuring and odometer error rectifying. An ad hoc wireless network is established base on the WLAN cards within the three robots [6]. Via this wireless ad hoc network, each robot can broadcast necessary information to all other robots. The experiments on these robots will verify the algorithm ’s efficiency in collaborating robot team using local communication.

　

　

Return

---