Error Aware Monocular Visual Odometry using Vertical Line Pairs for Small Robots in Urban Areas

We report a new error-aware monocular visual odometry method that only uses vertical lines, such as vertical edges of buildings and poles in urban areas as landmarks. Since vertical lines are easy to extract, insensitive to lighting conditions/shadows, and sensitive to robot movements on the ground plane, they are robust features if compared with regular point features or line features. We derive a recursive visual odometry method based on the vertical line pairs. We analyze how errors are propagated and introduced in the continuous odometry process by deriving the closed form representation of covariance matrix. We formulate the minimum variance ego-motion estimation problem and present a method that outputs weights for different vertical line pairs. The resulting visual odometry method is tested in physical experiments and compared with two existing methods that are based on point features and line features, respectively. The experiment results show that our method outperforms its two counterparts in robustness, accuracy, and speed. The relative errors of our method are less than 2% in experiments. Introduction We are interested in developing a visual odometry method for small robots in urban areas where tall buildings form a deep valley which can block GPS signals. Existing visual odometry methods are computationally challenging and cannot be used on small mobile robots with limited computation power. Employing a minimalist’s approach, we only focus on the robot ego-motion estimation on the ground plane using vertical lines under a regular pinhole camera due to common requirements and configurations of small robots. Building edges and poles are common features in urban areas (see Fig. 1(a)). These vertical lines are insensitive to lighting conditions and shadows. They are parallel to each other along the gravity direction. Extracting parallel lines using the gravity direction as a reference can be done quickly and accurately on low power computation platforms. Moreover, vertical lines are sensitive to robot motion on the ground plane. ∗This work was supported in part by the National Science Foundation under CAREER grant IIS-0643298 and MRI program under CNS-0923203. Copyright c © 2010, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. (a) -40 -20 0 20 0 50 100 150 200 250 300