Publications / 2020 Proceedings of the 37th ISARC, Kitakyushu, Japan

Research on Standardization of Construction Site Time-series Change Information as Learning Data for Automatic Generation of Work Plan of Construction Machinery in Earthworks

Takashi Otsuki, Hirokuni Morikawa, Yushi Shiiba, Seigo Ogata and Masaharu Moteki
Pages 1053-1060 (2020 Proceedings of the 37th ISARC, Kitakyushu, Japan, ISBN 978-952-94-3634-7, ISSN 2413-5844)

For earthwork, whether or not it will be possible to automatically generate a work plan for construction machinery has become the key to realizing autonomous construction of earthworks. The use of AI is being sought by some researchers for the automatic generation of construction setups. Efforts are being made to utilize the results of construction work plan of construction machinery carried out by skilled engineers as learning data when searching for rules to reproduce construction work plan of construction machinery. National Institute for Land and Infrastructure Management is considering acquiring work plans of construction machinery data in the MLIT ordering works, and to be going to provide these data. What is required for these data is to reproduce the history of construction progress and explain the reasons for deciding the construction work plan of construction machinery, and to provide those data in a format based on certain rules. We call this approach the examination of data standards for time-series change information at construction sites. As a starting point, we examined the acquisition method of the topographic shape and the effective display format. We tried the Voxel display as a data format that makes it easy to grasp the amount of construction progress and to add attribute information to each construction point. On the other hand, some experts have pointed out the usefulness of the point cloud data and the surface data generated by connecting the point cloud data in the design of the drainage slope. This paper reports the initiative.

Keywords: Work plan; Construction machinery; earthwork; Learning data; AI