Exploration of deep space regions such as the Jovian and Uranian systems，the Kuiper Belt，and even the outer reaches of the solar system is a key approach for humanity to expand the boundaries of cosmic cognition and to search for extraterrestrial life information. Therefore，it has become an important direction of global deep space exploration activities. Among them，landing and roving missions are the most direct and effective ways to explore deep space. The navigation，guidance，and control technologies for autonomous landing and roving in deep space are important research contents that urgently need to be broken through，and it is highly necessary to conduct a comprehensive review and summary of them. Therefore，this paper focuses on sorting out the research on navigation，guidance，and control for deep space landing and roving missions. It summarizes the classification and key technologies of autonomous navigation，classifies and elaborates on guidance and control technologies，and respectively analyzes their technical challenges and future development trends，so as to provide technical references for deep space exploration.

The Mars ascent vehicle is tasked with sending Martian samples into space orbit, and its trajectory optimization is a crucial part of the Mars sample return mission. The scenario of the Mars ascent mission is complex and subject to various uncertainties. However, traditional trajectory optimizations are mostly based on nominal models, with limited ability to address uncertainties. In this thesis, the uncertainties in the initial position and thrust amplitude were described as chance constraints, so a chance-constrained trajectory optimization problem was established. Then the optimal trajectory was iteratively solved using sequential convex optimization method. Numerical simulation results verified the effectiveness of the proposed method. Comparisons also demonstrated that this method achieved better performance indicators and exhibited less conservatism than traditional robust optimization. In addition, a comparative analysis of the risk management effect of chance constraints and the conservatism of probability approximation functions was conducted.

A rapid trajectory planning method of collision avoidance and obstacle avoidance is proposed to address the problem of landing on irregular small celestial body. An analytic form of landing trajectory is constructed using Bézier curve，and the collision avoidance and obstacle avoidance schemes are proposed. The guiding point of collision avoidance is set to prevent occlusion from the small celestial body，and the guiding point of obstacle avoidance is set to ensure landing along the normal direction of the landing plane. Considering the geometrical relationship between the lander and the small celestial body，and the monotonicity of height of the guidance point，constraints are established for the parameters of the collision avoidance point and the obstacle avoidance point. Under these constraints，the parameters are rapidly solved with the objective of combustion consumption. Additionally，the minimum landing time is investigated，with the thrust constraint determining the feasible range of landing time. Finally，the proposed trajectory planning method is simulated in a landing scenario of 433 Eros. Results show that the planned trajectory satisfies the constraints of collision avoidance，obstacle avoidance，and thrust control under various initial conditions，while maintaining high computational efficiency.

In response to the task requirements of high-precision real-time autonomous navigation for the asteroid probe during the descent and landing phase，considering the limited onboard computing resources and the complex terrain environment of the asteroid，this paper proposes a feature anchor point tracking-based autonomous landing navigation method. Firstly，the 3DSIFT algorithm is used to extract initial feature points from the three-dimensional point cloud data of the asteroid，and the spatial distribution feature selection mechanism is adopted to select feature anchors with high recognition degree，providing an accurate three-dimensional reference benchmark for subsequent visual tracking. On this basis，a lightweight convolutional neural network is used to directly detect and track the two-dimensional projections of the feature anchors in the image sequence，avoiding the computationally intensive feature matching steps in traditional methods，while ensuring navigation accuracy and significantly reducing computational complexity. Finally，based on the established 2D-3D point correspondence relationship，the asteroid lander’s pose parameters are accurately estimated using the EPnP method. Through software simulation and semi-physical experiments，it is verified that this method can effectively reduce the allowable computing power of the navigation system while ensuring the accuracy of pose estimation，providing an efficient and reliable solution for the autonomous navigation of the probe during the descent and landing phase of asteroid exploration.

In asteroid landing and exploration missions, the observation quality of optical navigation features is one of the key factors that influence the accuracy of autonomous navigation. To enhance the efficiency of navigation observation in asteroid landing missions, the observation requirements of navigation features were incorporated into the trajectory planning process, and a navigation observation-driven trajectory planning method for asteroid landing was proposed. The unevenly distribution of navigation features on asteroid surface was analyzed and a set of trajectory control points were determined by calculating their clustering centers. The landing trajectory was then redesigned by using Bezier Curve, and a two-dimensional waypoint sequence was obtained to fulfill the observation requirements. Furthermore, a three-dimensional navigation observation-driven trajectory was established by solving a trajectory optimization problem with waypoint constraints, revealing the impact of navigation feature distribution and observation requirements on landing trajectory. Numerical simulation results show that the proposed method can flexibly adjust landing trajectory and significantly improve the observation quality of navigation features.

The measurement reference bias caused by the relative installation error between landmark sensor and star sensor is one of the primary factors degrading navigation performance. In the space environment, positioning error of the landmark navigation system is increased due to slow time-varying measurement reference bias caused by deformation of sensors and their brackets. To cope with this problem, a parametric model for in-orbit calibration of time-varying measurement reference bias was proposed. Based on the model, a parameter estimation method for measurement reference bias was designed. Using this method, celestial landmarks and stars on the celestial sphere were observed with a landmark sensor and a star sensor respectively to obtain the line-of-sight (LOS) measurements of the landmarks in the inertial reference frame. Then the measurements were processed with a navigation filter to estimate the position and velocity of the spacecraft together with the measurement reference bias parameters in real time. Numerical simulations demonstrate that the proposed method effectively mitigates the unfavorable effect of measurement reference bias, significantly improving the positioning accuracy of the navigation system.

To address the design of transfer orbit from Distant Retrograde Orbit (DRO) to lunar circular orbit，a high-precision，computationally efficient and reliable method suitable for engineering applications was proposed. Based on a high-precision dynamic model，the impacts of orbital maneuver velocity increment，DRO orbital phase at maneuver time，and DRO selenocentric distance on the transfer orbit were quantitatively analyzed. Taking high-precision quantitative results as pre-generated empirical data and integrating them with analytical calculation，accurate computation of initial iteration values for design parameters was achieved，thus enabling the efficient solution of the transfer orbit design problem. Case study verification demonstrated that the proposed method was efficient and reliable with calculation completed within seconds. Quantitative analysis results and the proposed design method in this paper can provide technical reference for engineering applications.

In this paper, a human-machine collaborative obstacle avoidance control scheme for manned lunar lander GNC systems was proposed, built upon a framework of human-supervised autonomy. By constructing an obstacle avoidance safety point auxiliary selection algorithm, developing a hand-controlled semi-automatic obstacle avoidance control module, designing a human-controlled one-click emergency ascent scheme, and optimizing the human-computer interaction interface, astronauts’ autonomous decision-making support during the lunar landing process was realized. The results show that the system can effectively improve the reliability and safety of the lunar landing process and meet the needs of manned lunar exploration missions. This scheme provides a highly reliable and safe GNC system solution for manned lunar landing and descent missions, and has important engineering application value.

An adaptive selection method for aircraft landing safety zones was proposed. A two-level obstacle detection method was adopted, where the first level obstacle detection identified data deviating significantly from the base plane as obstacle area and non-obstacle areas were divided into sub-areas for second level obstacle detection. The partition size could be adaptively adjusted according to the proportion of obstacle areas. Taking into account factors such as slope difference with base plane, obstacle value, and number of obstacles, a multi-faceted obstacle degree evaluation indicator was proposed to complete the selection of obstacle avoidance safety zones. Experimental verification with laser point-cloud data shows that this method can identify obstacle terrain, and the selected obstacle avoidance safety zone has no obvious obstacles, indicating the effectiveness of the method.