Development of a persistent search algorithm using low cost microprocessor based robots as a test bench

Persistent searching is the continual search of a physical area using limited resources. Through the use of proven algorithms, coupled with the concepts involved with persistently searching an area, an algorithm has been developed. The algorithm provides for the persistent search of an area using three microprocessor based robot platforms. The system developed is a distributed one allowing for each robot platform to work in a group or individually as the case may be. It is assumed that the sensors used in the system are all homogeneous in nature but have a limited sensing range. A charge coupled device (CCD) camera system provides additional sensor information simulating either a Global Hawk or global positioning system (GPS) system allowing the robots to know their global location. The power system for each individual robot are also assumed to be homogeneous but have a limited capacity. Each robot has the capability to communicate with a central computer and to share information through the use of a radio frequency (RF) link. It is also assumed that there is no interference within the communication link between each of the robots. With these assumptions, the microprocessor based test bench uses low cost robots with infrared sensors and limited battery life to provide the ability to test the algorithms on actual hardware. This prevents the sole reliance on computer simulations. In each case the computer simulations will be considered the ideal situation. The success or failure of each individual hardware test is determined by how much deviation occurs between the actual hardware implementation and the computer simulation. The results showed that though the system does not perform perfectly or possibly as efficiently as it possibly could, it does work consistently on each map regardless of the maze configuration when the map is constrained. In the unconstrained case the system didn't seem to perform as well since the system had more open paths to choose from and therefore did not search the area very evenly.