面向月面复杂环境的自主导航及跨平台仿真系统的研究

目前，月面复杂场景下的无人车自主导航仍面临巨大挑战，研究面向月面非结构化环境的自主导航方法具有重要意义。。本文提出了一种基于地形约束的非结构化环境下的自主导航方法，采用一种基于八叉树地图的高效路径搜索和轨迹优化算法来生成轨迹，该算法可以有效地避开环境中的各种障碍物，到达指定的目的地。在Gazebo中进行了仿真实验，结果表明所提方法在无地图导航任务中取得了优异的性能。考虑到Gazebo在人机交互方面具有不足，进一步采用Windows中的Unity进行人机交互。在ROS与Windows交互方式上，设计了基于rosserial_serve的TCP连接，并采用Visual Studio解决方案生成Win32控制台应用程序，通过IO流读写本地文件的方式，完成目标点等信息的交互；同时结合PhysX物理引擎与人机交互的方式，提供了再现车辆在月面位姿、设置危险区域与规划路径点的功能，完成了基于Unity的寻路场景重现。

Research on Autonomous Navigation and Cross-platform Simulation System for Complex Environment on the Lunar Surface

The research on autonomous navigation of unmanned vehicles in lunar environments is still facing huge challenges and is of great importance in the scenario of lunar exploration. In this paper, an autonomous navigation method in unstructured environments based on terrain constraints is proposed. An efficient path search and trajectory optimization algorithm based on octree maps is proposed to generate trajectories that can effectively avoid various obstacles in the environment and reach the specified destination. The simulation experiments are conducted in Gazebo and the results show that the method proposed achieved excellent performance in mapless navigation tasks. However, Gazebo has deficiencies in human-computer interaction, the intention is to use Unity in Windows for human-computer interaction. In the ROS-Windows interaction method, a TCP connection based on rosserial_serve is designed and the Visual Studio solution is used to generate a Win32 console application that reads and writes local files through IO streams to complete the interaction of target points and other information; at the same time, the simulation system, which combines the PhysX physics engine with human-computer interaction, provides the ability to reproduce the vehicle''s position on the lunar surface, set hazard zones and plan path points, completing a Unity-based recreation of the pathfinding scenario.