Teleoperation of A Mobile Robot under Office Automation Floors with Visual Assistance

Human-robot collaboration has been discussed from various kinds of viewpoints such as human assistance, teleoperation, and cooperative behaviors[l]. While the cooperative behavior emphasizes the importance of interaction between the robot and human [2], the teleoperation emphasizes the importance of role assignment between the robot and human. In the teleoperation, the human should not operate the robot completely by hand, but the robot also should locally make decisions based on the facing environment. However, the human cannot perceive the situation of the remote workspace owing to the local observation. Therefore, the robot should communicate the facing situation to the human operator in order to reduce the perceptual gap between the robot and the human operator. In this paper, we focus on a mobile robot working under floors. Office automation (OA) floors are designed to wire various cables efficiently under floors. However, several panels must be removed when a cable is added to a network. This task takes much time and effort. In this paper, we develop a prototype of a mobile robot working under OA floor, and used a CCD camera as vision sensor to share the situation perception between a human and the robot as much as possible. The robot detects poles from the images, estimates the selflocation, and decides its moving direction. Various intelligent methodologies have been successfully applied to pattern recognition and motion planning problems [1, 3-5]. First, Kmeans algorithm is applied to extract possible areas of poles. Next, a steady-state genetic algorithm (SSGA)[6] is applied to detect poles from images by using a template matching method. Furthermore, we propose a method for estimating the self-location based on a series of visual perception, and teleoperation method of the robot. This paper is organized as follows. Section 2 explains the hardware of the robot, the pole detection method based on SSGA, and the control method of the robot. Section 3 shows that the robot can estimate the self-location and reach the target point through the teleoperation in several experimental results.