月球软着陆自主避障技术研究  

Research on Autonomous Obstacle Avoidance Technology forLunar Soft Landing

作　　者：高峰[1] 陈少松 牛津桥[2] GAO Feng;CHEN Shaosong;NIU Jinqiao(China Academy of Space Technology,Beijing 100094,China;PLA 63993,Beijing 101149,China)

机构地区：[1]中国空间技术研究院,北京100094 [2]中国人民解放军63993部队,北京101149

出　　处：《载人航天》2021年第4期403-411,共9页Manned Spaceflight

摘　　要：针对传统给定轨道和优化实现的软着陆制导方法对先验知识存在依赖,容易受建模偏差、不确定扰动等因素影响,且现有的障碍模型不具备较好的泛化能力等问题,提出了一种考虑多障碍约束下的燃料最优制导方法,并在凸锥模型的基础上提出了一种半球模型和通用凸包模型的建模方法。通过连续线性化实现了对非凸模型的凸化转化,将包含障碍约束的最优控制问题转化为等价的二阶锥优化问题。在障碍物分别为凸锥、半球和凸包的多障碍环境下对凸优化算法进行了验证。结果表明:本算法在3种障碍模型下均取得了较好的避障效果,其中凸包模型能够更准确地对障碍物进行近似。The traditional guidance method of soft landing based on given trajectory and optimization depends on the prior knowledge which is easily affected by modeling deviation,uncertainty disturbance and other factors.More over the existing obstacle models cannot generalize easily.To solve the above problems,an optimal guidance method considering multi-obstacle constraints was proposed,and a modeling method for hemisphere model and convex hull model based on convex cone model was innovatively proposed.Through successive linearization,the non-convex model was transformed into a convex model,and the optimal control problem with obstacle constraints was transformed into an equivalent second-order cone optimization problem.Finally,the convex optimization algorithm was verified in a multi-obstacle environment where the obstacles were convex cones,hemispheres,and convex hulls.The simulation results showed that the algorithm achieved excellent obstacle avoidance effects under the three obstacle models,and the convex hull model was more accurate and more universal in approximating obstacles.

关 键 词：月球软着陆 自主避障 轨迹优化 凸优化 

分 类 号：V448.2[航空宇航科学与技术—飞行器设计]