Sonar sensor virtualization for object detection and localization

Localization and obstacle detection have been unremittingly a challenging issue in the field of robotic from the past decades because of whimsical preciseness as well as cumulated drift of sensorial component and stochastic nature of robot's behavior in real time environment. The concentrated objective of this paper is to adopt the localization and object detection capability for the ground robot with the use of sonar array differently from the conventional sensors arrangement to utilize the magnitude of several sonar sensors and to preserve the port and weight constraint of respective robotic platform. The overall localization and object detection experiment is based on the small scale robotics and local spatial information which has been accomplished in a 2D geometrical grid map. For localizing the odometry data directly fetched from the servo motor has been used and for the object detection we have used sonar array because of its cost effectiveness and low rate of interference and sensitivity to the electromagnetic field effect. As both of the localization and object detection are not completely error free in any robotic platform, we have employed mathematical approach and well defined algorithm to reduce the error and inaccuracy. For localizing the objects in the map, a trigonometric approach known as triangulation has been employed. The use of three sonar sensors in an innovative way to acquire the virtualization of twelve surrounding sonar sensors and the integration of trigonometric and geometric approaches have made it possible to localize mostly all of the objects around the robot and enabled the robot to identify the obstacles throughout the trajectory of its movement. The implementation of our method has been performed on the ground robot built using LEGO Mindstorm EV3 intelligent brick and LEJOS package have been used for programming.