Commission V, WG V/4 KEY WORDS:as-built data, terrestrial laser scanner, panoramic images, CAD models, industrial applications, large scale scanning surveys
ABSTRACT:As-built CAD data reconstructed from Terrestrial Laser Scanner (TLS) data are used for more than two decades by Electricité de France (EDF) to prepare maintenance operations in its facilities. But today, the big picture is renewed: "as-built virtual reality" must address a huge scale-up to provide data to an increasing number of applications. In this paper, we first present a wide multi-sensor multi-purpose scanning campaign performed in a 10 floor building of a power plant in 2013: 1083 TLS stations (about 40.10 9 3D points referenced under a 2 cm tolerance) and 1025 RGB panoramic images (340.10 6 pixels per point of view). As expected, this very large survey of high precision measurements in a complex environment stressed sensors and tools that were developed for more favourable conditions and smaller data sets. The whole survey process (tools and methods used from acquisition and processing to CAD reconstruction) underwent a detailed follow-up in order to state on the locks to a possible generalization to other buildings. Based on these recent feedbacks, we have highlighted some of these current bottlenecks in this paper: sensors denoising, automation in processes, data validation tools improvements, standardization of formats and (meta-) data structures. * Corresponding author.
CONTEXT AND CONTRIBUTIONSIndustrial installations need constant maintenance operations to keep performance and safety at their highest levels. In the last decades, as-built data (Terrestrial Laser Scanner data -TLS-, asbuilt CAD models reconstructed form scans, panoramic images) have proven their efficiency for preparing decision and executing several maintenance operations. But today, the big picture is renewed: "as-built virtual reality" must address a huge scale-up to provide all the current and incoming applications that rely on these data sets. For example, at Electricité de France (EDF), owner and operator in France of 58 nuclear power stations, many use cases have been recently developed for maintenance of the plants using as-built virtual reality. From a few areas, many buildings will soon be digitized. Surveyors on field switch from "modest" surveys with 50-100 stations to very large surveys with at least 1000 stations. Computers, software, storage and networks change from handling Megabytes to hundreds of Terabytes of data: this is no more business as usual.In this context, through this paper, our contributions are the following: depicting the current trends of industrial close-range as-built data creation for industrial uses at EDF, by describing real cases, scales and issues, sharing feedbacks from a first wide multi-data closerange acquisition campaign (1083 TLS stations, 1025High Resolution panoramas, 2D maps and 3D models as-built reconstruction and processing), highlighting short and medium terms challenges of close-range acquisition tools and methods...