Abstract:
Designing and fabrication of mobile security platform is a relatively new development in
security conscious world. These mobile security platforms are being developed keeping in view
their employment environment. In general, these platforms perform two main tasks, first is the
basic task of mapping the environment in which it will operate as well as self localization and
thereafter navigation in that particular environment, second is the specific security task of
surveillance. Mapping and localization has become one of the mainstream research areas in
mobile robotics and is generally performed using simultaneous localization & mapping (SLAM)
algorithms. These SLAM algorithm are broadly classified in three categories, first is based on
calculation methods like Kalman Filter or the particle filter, hybrid comprising of both filters and
graph-based, second is based on sensors like vision, range measurement devices and odometry,
third is based on structure like online SLAM and full SLAM. Surveillance task is performed by
assessment algorithms supported by different sensors like acoustic, vibration, passive infrared,
microwave, optical, ultrasonic and vision.
This work focuses on mapping and localization part of a mobile security platform which has
to be employed in an unknown indoor environment. In this context, an in-depth study has been
carried out on SLAM algorithms developed to-date along with different sensors available. Visual
SLAM algorithm is tested with latest ZED stereo camera in an unknown indoor environment for
its use in mobile security platform.
