A REAL-TIME AUTONOMOUS FACE-TRACKING SYSTEM BASED ON A 2-DOF ARTICULATED MANIPULATOR PLATFORM USING EXTENDED KALMAN FILTER
There is an increasing demand for autonomous tracking applications in the industrial context which ranges from driver monitoring in semi-autonomous vehicles to human-robot interaction (HRI) to facilitate situational awareness in collaborative robots. In order to address the same, a system to track the human face in real-time has been developed and the system is capable of moves accordingly so that the face always remains in the range of visibility of the autonomous system. The system consists of open-source hardware and software to design the Tracking Algorithms which utilizes the Extended Kalman Filters (EKF) at its core. In addition to the basic model, this paper uses a hybrid model, implemented using both Extended Kalman Filters and Viola Jones in conjunction with Iterative Learning Control (ILC) intelligent tuning of PID loop. Performance evaluation of the system has been done in Solidworks and MATLAB. The proposed model with two different control methodologies along with the modified Extended Kalman Filter and Viola Jones Based algorithm has a shorter delay time and produced stable responses over traditional viola jones wavelets based approach.
Object tracking, Face Detection, Robots, Autonomous Systems, Control System, Kalman Filter.