Powered-descent landing GNC system design and flight results for Tianwen-1 mission

The powered-descent landing (PDL) phase of the Tianwen-1 mission began with composite backshell—parachute (CBP) separation and ended with landing-rover touchdown. The main tasks of this phase were to reduce the velocity of the lander, perform the avoidance maneuver, and guarantee a soft touchdown. The PDL phase overcame many challenges: performing the divert maneuver to avoid collision with the CBP while simultaneously avoiding large-scale hazards; slowing the descent from approximately 95 to 0 m/s; performing the precise hazard-avoidance maneuver; and placing the lander gently and safely on the surface of Mars. The architecture and algorithms of the guidance, navigation, and control system for the PDL phase were designed; its execution resulted in Tianwen-1’s successful touchdown in the morning of 15 May 2021. Consequently, the Tianwen-1 mission achieved a historic autonomous landing with simultaneous hazard and CBP avoidance.

Performance Analysis of Mars-Powered Descent-Based Landing in a Constrained Optimization Control Framework

It is imperative to find new places other than Earth for the survival of human beings. Mars could be the alternative to Earth in the future for us to live. In this context, many missions have been performed to examine the planet Mars. For such missions, planetary precision landing is a major challenge for the precise landing on Mars. Mars landing consists of different phases (hypersonic entry, parachute descent, terminal descent comprising gravity turn, and powered descent). However, the focus of this work is the powered descent phase of landing. Firstly, the main objective of this study is to minimize the landing error during the powered descend landing phase. The second objective involves constrained optimization in a predictive control framework for landing at non-cooperative sites. Different control algorithms like PID and LQR have been developed for the stated problem; however, the predictive control algorithm with constraint handling’s ability has not been explored much. This research discusses the Model Predictive Control algorithm for the powered descent phase of landing. Model Predictive Control (MPC) considers input/output constraints in the calculation of the control law and thus it is very useful for the stated problem as shown in the results. The main novelty of this work is the implementation of Explicit MPC, which gives comparatively less computational time than MPC. A comparison is done among MPC variants in terms of feasibility, constraints handling, and computational time. Moreover, other conventional control algorithms like PID and LQR are compared with the proposed predictive algorithm. These control algorithms are implemented on quadrotor UAV (which emulates the dynamics of a planetary lander) to verify the feasibility through simulations in MATLAB.

Improving the efficiency of reinforcement learning for a spacecraft powered descent with Q-learning

Reinforcement learning entails many intuitive and useful approaches to solving various problems. Its main premise is to learn how to complete tasks by interacting with the environment and observing which actions are more optimal with respect to a reward signal. Methods from reinforcement learning have long been applied in aerospace and have more recently seen renewed interest in space applications. Problems in spacecraft control can benefit from the use of intelligent techniques when faced with significant uncertainties—as is common for space environments. Solving these control problems using reinforcement learning remains a challenge partly due to long training times and sensitivity in performance to hyperparameters which require careful tuning. In this work we seek to address both issues for a sample spacecraft control problem. To reduce training times compared to other approaches, we simplify the problem by discretising the action space and use a data-efficient algorithm to train the agent. Furthermore, we employ an automated approach to hyperparameter selection which optimises for a specified performance metric. Our approach is tested on a 3-DOF powered descent problem with uncertainties in the initial conditions. We run experiments with two different problem formulations—using a ‘shaped’ state representation to guide the agent and also a ‘raw’ state representation with unprocessed values of position, velocity and mass. The results show that an agent can learn a near-optimal policy efficiently by appropriately defining the action-space and state-space. Using the raw state representation led to ‘reward-hacking’ and poor performance, which highlights the importance of the problem and state-space formulation in successfully training reinforcement learning agents. In addition, we show that the optimal hyperparameters can vary significantly based on the choice of loss function. Using two sets of hyperparameters optimised for different loss functions, we demonstrate that in both cases the agent can find near-optimal policies with comparable performance to previously applied methods.

Guidance Navigation and Control for Chang’E-5 Powered Descent

To achieve the goal of collecting lunar samples and return to the Earth for the Chang’E-5 spacecraft, the lander and ascender module (LAM) of the Chang’E-5 spacecraft successfully landed on the lunar surface on 1 Dec., 2020. The guidance, navigation, and control (GNC) system is one of the critical systems to perform this task. The GNC system of previous missions, Chang’E-3 and Chang’E-4, provides the baseline design for the Chang’E-5 LAM, and the new characteristics of the LAM, like larger mass and liquid sloshing, also bring new challenges for the GNC design. The GNC design for the descent and landing is presented in this paper. The guidance methods implemented in the powered descent are presented in detail for each phase. Propellant consumption and hazard avoidance should be particularly considered in the design. A reconfigurable attitude control is adopted which consists of the quaternion partition control, phase and gain stabilization filter, and dual observer. This controller could provide fast attitude maneuver and better system robustness. For the navigation, an intelligent heterogeneous sensor data fusion method is presented, and it is applied for the inertial measurement unit and velocimeter data. Finally, the flight results of the LAM are shown. Navigation sensors were able to provide valid measurement data during descent, and the thrusters and the main engine operated well as expected. Therefore, a successful soft lunar landing was achieved by the LAM.