Structure recognition using the machine learning for accident risk reduction in the decommissioning

This paper describes fundamental examinations on the structures discriminations utilizing machine learning approach. In the decommissioning work, troubles during the task execution by remotely operated robots had sometimes occurred. In order to decrease such risks, it is required to install the functions for the improvement of the environment awareness to the operation system. We are motivated to introduce machine learning technique to discriminate the structures in the plant based on sensory data measured by the robot. In this paper, we report experimented results of the structures discrimination based on point cloud data obtained by 3D laser scanner.