Improving the Computational Efficiency of ROVIO

ROVIO is one of the state-of-the-art monocular visual inertial odometry algorithms. It uses an Iterative Extended Kalman Filter (IEKF) to align features and update vehicle state simultaneously by including feature locations in vector IEKF. This algorithm single-core intensive, which allows other cores be used for algorithms, such as object detection path optimization. However, computational cost grows rapidly with maximum number track. Each adds three new states (a 2D bearing inverse depth), leading bigger matrix multiplications that are computationally expensive. The main load iterative step In this work, we reduce average [Formula: see text] on NVIDIA Jetson TX2, without affecting accuracy algorithm. gain mainly achieved exploiting sparse matrices ROVIO. Furthermore, peaks pre-selecting based their already calculated FAST score. combination both modifications us run restricted Raspberry Pi Zero 2[Formula: text]W.