Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : May 27, 2020 - May 30, 2020
These days, the number of people who need mobile manipulator-type human support robots has been increasing. Although in conventional navigation, these robots move only in free areas where no object is placed, there are many objects in human living environments, making it difficult for them to move efficiently. Therefore, they are required to go to the goal while a movable object to clear the passage, by adding the movability of objects to the environment maps. In this study, we propose topological navigation that uses a topological map based on sensing information and representing a path that robots can pass, and a metric map based on geometrical information of objects and representing the presence of objects. In the proposed navigation method, robots can determine whether an object on the passage and can select the most efficient path that minimizes the move cost. We confirmed the validation of the proposed method by experiments on real robots.
