UC Riverside

This paper presents the problem of finding the optimal rendezvous point that minimizes the sum of squared distances to it from two or more preset convex regions (circles or ellipses) using a set of mobile robots. Each robot only has information of its own region, its own position, its neighbor(s), and their position(s). We employ the distributed shortest-distance algorithm presented in [1] that minimizes one's distance to its region while moving towards the direction of its neighbor(s). This algorithm is implemented in an infrastructure built using Advanced Robot Interface for Application (ARIA), which provides various tools to communicate and network in C++. This framework is used to program a set of commercially available robots called iRobot Create. Each robot is equipped with a Bluetooth module to send encoder data and receive control commands to and from a central computer. A downward-looking web camera is affixed to the ceiling to emulate proprioceptive and exteroceptive sensor data for each robot. The computer, using the camera, runs ARToolkit to detect the pose of unique markers mounted on each robot. The control data is calculated for each robot using self and neighbor position data and is sent back to each robot from the central computer. We further improve the algorithm by introducing potential functions in the control law to enable collision avoidance among robots.