Sun Observation Image Based Celestial Navigation For Lunar Rover

In order to carry out reconnaissance survey and complete scientific detection tasks under complex environment on the moon, lunar rover require the support of navigation systems. To ensure that it can execute survey missions correctly under long time, long range conditions, its navigation system must have a high degree of autonomy, especially in high-precision positioning and orientation capabilities. Navigation methods that could be used for lunar rover include inertial navigation, visual navigation and celestial navigation and so on. Autonomous celestial navigation is a kind of error-bounded absolute navigation, and can correct accumulated error of relative navigation (such as inertial navigation and visual navigation). On the moon, other orientation methods are all ineffective, only celestial orientation can provide absolute heading information for the lunar rover. In our research about sun observation image based celestial navigation for lunar rover, the main innovations are as follows:1. High-performance sun image centroid extraction algorithm in sun sensor.Sun sensor's output is one of the major source information for lunar rover celestial navigation. In this special environment on the moon, to weaken the impact of lunar dust on the imaging, we can use optical lens camera in visual navigation as sun sensor. In order to reduce the image noise and typical image degradation in sun sensor, this paper proposes a novel high resolution Zernike moments based centroid algorithm. The edge is located by Sobel operator, the edge is then relocated with sub-pixel accuracy by Zernike moment operator and the image centroid is obtained using least square method. Also degraded images are processed with the proposed valid edge point detection method before using the Zernike moment based method. From theoretical analysis we know that Zernike moment edge detection improves the accuracy of circle fitting method. In simulation and experiment, this method is compared with traditional centroid method, threshold based centroid method and the circle fitting method. The results show that this method has higher accuracy and higher stability, and it can play an important role to improve lunar rover's navigation accuracy.2. A non-iterative celestial positioning method for lunar roverThe traditional celestial positioning method for lunar rover is the iterative analytical altitude difference method (the iterative algorithm), which is essentially a nonlinear least squares estimation algorithm, requires the initial position value in the iterative solving process until convergence. Aiming at the shortcomings for the iterative method, we propose a novel non-iterative method for lunar rover positioning. This method can be used to locate the rover without initial position information and without an iterative process. In addition, to accelerate convergence it can also used as the initial value of iterative algorithm. Compared with spherical geometry used in the iterative algorithm, analytic geometry is mainly used in non-iterative algorithm. This method saves computing time, has higher precision orientation, but positioning is still greatly affected by noise. Improvements in noise suppression have been made in Part 3 below.3 Simultaneous celestial positioning and orientation for the lunar roverThis paper focuses on simultaneous celestial positioning and orientation for the stationary lunar rover, the method without iterations and can suppress noise effect. Firstly we measure the sun orientation by a sun sensor and the gravity orientation by an inclinometer; secondly combine them with sun ephemeris; finally solve this problem with quaternion estimation (QUEST). By analyzing affects of the sensor's system error (such as the installation errors, bias, etc.) and random noise under stationary conditions, these two types of errors can be eliminated using joint calibration. Executing the SCPO on the moon is demonstrated through simulation. The algorithm is carried out with an experiment on the earth surface, and results demonstrate the high consistency of our simulation.4 Angular Velocity Current Statistical Model based Continuous Celestial Navigation for Lunar Rover The key point of lunar rover's long time, long range autonomous navigation is to acquire state information accurately and continuously within the process of moving, and to overcome shortcomings of celestial orientation under the parking conditions, A novel celestial navigation method is proposed to solve this problem, which is based on celestial observation and output of angular rate gyroscope. In this paper, the star's direction vector from star sensor is treated as observation for system measurement equation; kinematics model is established using this angular velocity current statistical model (ACSM) for the rover, and its attitude quaternion differential equation is treated as system state equation. Finally, the extended Kalman filter (EKF) gives the solutions of rover's position, heading and attitude. Simulation results show that this method could obtain higher celestial navigation accuracy in real-time.