Simultaneous localization and mapping on an FPGA based robot with stereo vision

Location and map information is an important foctor affecting the success of a mobile robot. In this study, it is aimed to construct an FPGA based architecture for Extended Kalman Filter assuring simultaneous localization and mapping on a mobile robot with stereo cameras. In the system, corners in the images are regarded as feature points. The positions of the corners with respect to the robot are computed by applying stereo matching and location measurement in 3D. The 3D positions of the corners are fed into the Extended Kalman Filter to be used in the measurement update. The location change in the actuators of the robot is used in motion update.The study is realized on a Virtex-5 XC5VLX50 FPGA core from Xilinx. OmniVison OV7720 digital cameras are used in order to sense the images. MT4HTF3264HY-53E DDR2 SODIMM is used to store the images.