Fuzzy Rule Interpolation Based Object Tracking and Navigation for Social Robot

The human-robot interaction will be more and more important in the close future. The behavior based robot control is simplified using fuzzy control. If the robot is doing a complex task, its behavior can be described with fuzzy rules. This kind of control method can be is easier implemented then the classical ones. Using fuzzy rule interpolation when the number of rules is high, the system can be described by the significant rules only. The paper presents a robot with ball playing task. The ball is detected using image processing methods. The images are processed using OpenCV library. The image processing function is implemented as a task of ROS node (Robot Operating System) on a Raspberry Pi computer placed on the robot. The mobile robot moves in holonomic way. In the field augmented reality markers are used for localization and navigation. The markers are detected on the same camera image as the ball. The markers positions are known by the robot. The robot computes its own position according to the detected markers. The navigation control is based on fuzzy rule interpolation, in this way the robot can avoid obstacles and approach the destination point.