A study on mapping of multi-modal environment for mobile robot in semi-dynamic environment

为了探索移动机器人在半动态环境下的多模态映射技术，我们开发了三个子系统，(1)利用彩色图像数据和距离数据提高物体姿态精度的子系统，(2)在环境中估计导师在教学物体时手指指向手势的方向的子系统，以及(3)教授物体的语义信息的子系统，如名称和所有者。教师和机器人之间使用语音交互。我们验证了这些子系统。

In order to explore multimodal mapping techniques for mobile robot to work in semi-dynamic environment, we have developed three subsystems, (1) a subsystem to improve accuracy of object's pose by using color image data in addition to its range data, (2) a subsystem to estimate direction of finger pointing gesture of a tutor in teaching objects in the environment, and (3) a subsystem to teach semantic information of the objects, such as names and owners, ususing voice interaction between a tutor and a robot. And we validated these subsystems.