Research On Autonomous Optical Navigation For Pinpoint Lunar Soft Landing

The lunar landing exploration which can achieve more science values, and it is one of the most important lunar exploration method. Due to the signal delay produced by the large distance between the earth and lunar, and the short dura-tion of descent, the spacecraft must have the ability of autonomous guidance, navigation and control, which is a critical technology for deep space exploration mission. With the supports of the National Natural Science Foundation of China'Theory and Method of Deep Space Autonomous Navigation', this dissertation deeply researched the autonomous navigation methods for autonomous pinpoint lunar landing. The main contents and innovations of this dissertation are as fol-lows:Firstly, the method of autonomous vision feature detection methods and safe landing point selection methods are deeply researched. Crater and rock are the most notable terrain features on lunar surface, which can be detected and tracked as navigation landmark, on the other hand, as the main hazard terrain features which must be detected and avoidance. In this chapter, based on passive surface image, the autonomous crater and rock detection methods are proposed respec-tively. As maneuver ability is one of the most important factors for safe landing point selecting, the analytic expression of maneuver capability is derived to as-sessment the maneuver capability in real time. In the last, the fuzzy rules are used to fuse the hazard detection result, maneuver capability, the distance to nearest hazards and the distance to science interesting to comprehensive evaluation the safeness of terrain and select the best safe landing point.Secondly, the crater landmark tracking based navigation for pinpoint lunar soft landing is researched. During the soft landing, due to the change of space-craft motion states and visual angle, the crater image rotation, size and shape can be greatly changed, and the crater is difficulty to be matched by general image match methods. In this paper, the affine invariants of crater are advanced to match the detected crater with crater database, which can effective determine cra-ter's global position on lunar surface. For the measurements of crater's line of sights, the Unscented Kalman Filter is used to estimates spacecraft position and attitude. In addition, the propagating covariance in computer vision is sued to analysis the states estimation error, and the crater landmark selecting strategy is given.Thirdly, the digital elevation map matching based navigation method for lu-nar soft landing is proposed. For the real time 3D data of lunar surface which can be obtain by LIDAR, the local covariance is calculated, and the local maximum covariance points in global terrain are searched as the 3D terrain features. Then, based on the geometry invariants of relative position and angle, the voting strat-egy is used to match the real time 3D features with global feature database, and the global position on lunar surface can be determined. As the unknown noise statistical of 3D features'position in spacecraft body coordinate system, the Sage-Husa noise estimator is combined with the Adaptive Iterated Kalman Filter to estimate the noise and spacecraft motion states, which can reduce the bias and the estimation error effectively by increasing only a few simple iterative opera-tions.Lastly, the multi-measurements fusion based navigation for lunar soft land-ing is researched. During the final landing phase, the crater landmark may can not matched due to the large change of resolution. In this paper, the relative measure model of vision features in sequence images is derived, and the Kalman Filter is used to integrate the measurements of landmark's line of sight and rela-tive measurements in image sequence to capture the advantages of both meas-urements. For the large error of inertial navigation and the shortage of substantial burden and slow data update rates in vision based navigation, the Extend Klaman Filter is used to integration the measurement of IMU, image coordinates of landmarks and opportunistic features to estimate spacecraft's 6Dof motion pa-rameters and vision features'position in inertial reference frame. The proposed scheme can capture both the advantages and greatly improve the navigation pre-cision.