Attitude/Position Estimation of Rigid-Body using Inertial and Vision Sensors

Abstract

This paper is concerned with the attitude/position estimation of a rigid-body using inertial and vision sensors. By employing the Newton-Euler method, a kinematic model is developed for the rigid-body by treating the inertial measurements as inputs. Based on the coordinate transformation, a nonlinear visual observation model is proposed by using the image coordinates of feature points as observations. Then the Extended Kalman filter (EKF) is utilized to estimate the attitude/position recursively. The effectiveness of the proposed algorithm is evaluated by simulation.