Visual Odometry analysis of the NASA Mars 2020 Perseverance rover's images

Visual Odometry is a key technique used by planetary exploration rovers to achieve high localization accuracies during the motion. Visual Odometry methods allow to reconstruct the position and attitude of moving assets by processing visual inputs acquired by the onboard stereocameras. Vision-based localization techniques have supported the operations of planetary rovers during the last two decades, including the NASA's Mars 2020 mission with the Perseverance rover. Our work is focused on the analysis of the stereo-pairs acquired by Perseverance's navigation cameras on sol 65 to retrieve the vehicle's position and attitude along the path. The images are processed by using a 3D-to-3D stereo vision-based algorithm that employs the CAHVORE camera model to account for nonlinear optical effects characterizing the raw images. By comparing the triangulated coordinates of the same set of landmarks observed before and after the entire drive, we show that the trajectory retrieved with our Visual Odometry software improves the archived rover's position.