基于四叉树直方图的空间关系描述理论与机器人问路导航方法研究

Any space-related intelligence system must have the ability to describe the spatial relations,so how to achieve the spatial relation description is an important research topic. Mobile robots need to realize the natural langural communication with human on the path of travel of the space,so the robot navigation by asking the way is one of the main content of the robot navigation research.Firstly, the project team by theme cognitive experiments further improves the theoretical framework of spatial directional relation, proposes quadtree histogram parameter optimization algorithm and improves the formal model of spatial directional relation system;Secondly,by in-depth analysis of the human characteristics of natural language to describe the complex spatial relations and dynamic spatial relations, the project team proposes the spatial relation natural language description methods of the detail spatial relation and the dynamic spatial relation; Thirdly, the project team proposes dynamic spatial orientation method based on histogram template by taking into account the characteristics of dynamic spatial orientation; Finally, the project team proposes the robot navigation method by asking the way, and designs a robot navigation system based on robot NAO.The pre-research outcome of the project team proved the feasibility and effectiveness of the project. The project adapts to the development of artificial intelligence and image understanding, and provides strong support for the barrier-free communication between machines and human on the concept of spatial relation.

任何与空间有关的智能系统都要有能力对空间关系进行描述，因此人工智能研究的一个重要任务就是实现空间关系描述。针对此目标，本课题对基于四叉树直方图的空间关系描述与机器人问路导航方法进行研究。本项目首先通过主题认知实验进一步完善空间方向关系的基础理论框架，基于已有的四叉树直方图模型提出四叉树直方图参数优化算法，完善空间方向关系形式化模型体系；然后通过深入分析人类描述复杂空间关系以及动态空间关系的自然语言特点，提出细节空间关系以及动态空间关系自然语言描述方法；接着考虑到动态空间定位的特点提出基于直方图模板的动态空间定位方法；最后提出交互式机器人问路导航方法，并在机器人NAO平台上构建基于动态空间关系描述的机器人问路导航系统。本项目组已经取得的预研成果证明了本项目的可行性和有效性，适应了人工智能以及图像理解的发展要求，为使机器与人在空间关系概念上实现无障碍的沟通提供了保障。