Optimal Soft Landing For Extraterrestrial Considering Complex And Multi-obstacle Constraints

With the innovation and development of aerospace technology,China has made remarkable achievements in many fields such as space exploration in recent years.Breakthroughs such as the operation of the space station,the successful conclusion of the lunar exploration project,the successful landing of Tianwen-1 on Mars,and the opening of Mars inspection by Zhurong are milestones in the development of China’s aerospace technology.With the accumulation and innovation of technology,future exploration needs are bound to target further deep space,more dangerous polar regions,and more mysterious planets,followed by more unknown,complex,uncertain,and multi disturbance extreme challenges.As an important part of the exploration mission,soft landing technology is the premise of star catalog exploration.Its implementation effect directly determines the success of the mission,and there are many disturbances and uncertainties in the whole process.Therefore,this paper proposes an optimal soft landing guidance technology for extraterrestrial objects considering complex multi obstacle constraints,to provide safe soft landing support for the detection needs of the extreme environment in the new era.In order to achieve a successful soft landing in the extraterrestrial complex environment,the moon lander needs to have autonomous trajectory planning and obstacle avoidance technology with high accuracy and safety assurance.For the traditional direct method of solving the soft landing trajectory based on optimal control,due to the introduction of complex and uncertain obstacle constraints,the optimal control problem will be strongly non-convex and ill-conditioned.In addition,the existing obstacle models can not generalize easily.In response to the above problems,this paper proposes an optimal guidance method considering multi-obstacle constraints,and innovatively proposes a modeling method of convex hull obstacle constraint.Through successive linearization,the non-convex model is transformed into a convex model,and transformed into an equivalent second-order cone optimization problem.Finally,the simulation results are given to verify the feasibility of the convex hull obstacle model and the effectiveness of the convex optimization technology.The direct modeling of obstacles in traditional obstacle avoidance methods will bring non-convexity to the trajectory optimization problem.The convexized model will lose a certain degree of accuracy,and the obstacle model is not universal.In response to the above problems,this paper proposes an improved relatively safe flight path method sui Tab for the second-order cone optimization problem,which can generate a safe convex hull flight path that satisfies the obstacle avoidance condition,and the constraint is convex;the jumping point method is improved to merge The number of adjacent nodes is reduced,and a deceleration time allocation method sui Tab for planetary soft landing is proposed.The method proposed in this paper avoids the direct modeling of obstacles and is sui Tab for any type of planetary obstacle avoidance.To effectively deal with the impact of multi-source disturbance and improve the online decision-making ability of the lander,this paper explores and studies its feasibility in the soft landing mission based on artificial intelligence technology.Combined with the mission environment and platform model,a planetary soft landing control algorithm based on reinforcement learning is designed,and the problem of sparse reward is solved by introducing reference speed,The fuel consumption and attitude are introduced into the reward design to realize fuel optimization and attitude constraints.Finally,the classical algorithms DDPG,TD3,and SAC in the field of reinforcement learning are used for training and testing respectively.The results prove the advantages of the intelligent algorithm in learning efficiency and accuracy and provide a reference for the subsequent practical engineering implementation.