基于场景图谱的室内移动机器人目标搜索

在移动机器人执行日常家庭任务时，首先需要其能够在环境中避开障碍物，自主地寻找到房间中的物体。针对移动机器人如何有效在室内环境下对目标物体进行搜索的问题，提出了一种基于场景图谱的室内移动机器人目标搜索，其框架结合了导航地图、语义地图和语义关系图谱。在导航地图的基础上建立了包含地标物体位置信息的语义地图，机器人可以轻松对地标物体进行寻找。对于动态的物体，机器人根据语义关系图中物体之间的并发关系，优先到关系强度比较高的地标物体旁寻找。通过物理实验展示了机器人在语义地图和语义关系图的帮助下可以实现在室内环境下有效地寻找到目标，并显著地减少了搜索的路径长度，证明了该方法的有效性。

Indoor mobile robot target search based on the scene graphs

When a mobile robot performs daily household tasks, it must avoid obstacles in the environment and find objects in the room autonomously. To enable mobile robots to effectively search for target objects in indoor environments, an indoor mobile robot target search method based on a scene graph is proposed. Its framework combines a navigation map, a semantic map, and a semantic relationship graph. Based on the navigation map, a semantic map containing the location information of the landmark objects is established, so the robot can easily search for these objects. For dynamic objects, the robot will first look for landmark objects with a higher relationship strength based on the concurrent relationship between the objects in the semantic relationship graph. Physical experiments show that the robot can effectively find the target in an indoor environment using the semantic map and the semantic relationship graph, significantly reducing the search path length and hence proving the effectiveness of the method.