The ever growing challenges in hostile environments, health care and warzone require accurate indoor localization and surveillance. The de facto localization technique using GPS has poor indoor performance due to the complexity of the indoor environment. Other Radio frequency based indoor localization techniques are unable of accurate localization due to multipath fading.
In this thesis, a system consisting of Cricket wireless sensor motes, a camera and a Pan/Tilt gimbal is proposed to solve the indoor localization and surveillance problems. The system is easy to deploy, is cost effective and gives accurate results. The Crickets motes use the Time Difference of Arrival (TDoA) between the RF and the ultrasound signals to estimate the distance of the object. Multilateration is used to calculate the position of the object in the reference beacon coordinate system. This position is then transformed to the object coordinate system to calculate the pan and tilt angles of the gimbal which are then used to direct the camera to the object. The programming language JAVA was used to develop a GUI program to interface the gimbal, the camera and the Cricket motes.
The localization and tracking of the object was successfully carried out in the laboratory. The accuracy of the system was tested using a laser pointer mounted on top of the camera and was shown that the system tracked the object with negligible error.