Initial position and attitude determination of lunar rovers by INS/CNS integration

Navigation methods of lunar rovers like inertial navigation system (INS), dead reckoning and visual odometry must be provided with the initial navigation parameters such as initial position and attitude. The accuracy of these initial navigation parameters has a significant impact on the overall navigation accuracy. Taking INS as an example, on the Earth, initial position can be obtained from global positioning system (GPS) or other ground facilities, and initial attitude can be obtained through initial alignment. However, there is no GPS on the lunar surface, and the lunar rotation rate is too small to execute alignment. For solving this problem, a new initial position and attitude determination method based on INS/CNS (celestial navigation system) integration is presented in this paper. Star altitude error caused by the biases of accelerometers is considered and its corresponding measurement equation is established accurately for the first time. The horizontal velocity errors, starlight vectors and star altitudes are used to estimate the initial position and attitude by an unscented Kalman filter (UKF). Furthermore, the advantage of this new method is that the INS sensors errors can be estimated accurately. Semi-physical experiments show that higher estimation accuracy is achieved by this new method compared with that of the traditional INS alignment method and INS/CNS initialization methods. These results demonstrate that it is a promising and attractive method to provide the initial position and attitude for lunar rovers.