Self-localization method using a single omni-directional camera based on landmark positions and arrangement

This paper proposes a self localization method for autonomous vehicles with just a omni-directional camera. Proposed method uses positions and arrangement of landmarks which are preliminarily obtained, and estimates the position of omni-directional camera based on the cosine formula. Also measurements of distances between the camera and the landmarks are not required, which are needed for conventional self-localization method but becomes the major reason of estimation errors. So, this method can obtain a large advantage to avoid the estimation errors due to the measurement error of the distance between camera and landmarks. The geometric nonlinear simultaneous equations consists of the cosine formulas are solved by recursive least squares method. Performance and accuracy are indicated by the results of simulations.