3DIMpact: Research, enabling, outreach and CPD

3DIMPact carries out a wide variety of scientific and applied research in collaboration with other UCL departments as well as external organisations and leading industry experts.

Robotics sensing, testing and teleoperations

UCL 3DIMPact is part of UCL Robotics. UCL Robotics is a cross-faculty initiative to develop a world-class teaching and research platform in Robotics and Autonomous Systems. We cover applications in the Life Sciences, Healthcare, Manufacturing and Creative Industries.

As part of an EPSRC project into remote dexterous manipulation, UCL researchers at the Department of Computer Science and Civil, Environmental and Geomatic Engineering are developing one of the longest anthropomorphic dexterous snake-like robot. Capable of moving into difficult and inaccessible places, we demonstrated the robots ability to perform dexterous hand like tasks across large distances. In paying homage to the project’s inspiration Mr Tickle from the Mr Men children’s books, we used the unique robot to extend our arms reach and perform everyday tasks. The robot consists of two unique robots supported by UK companies ShadowHand and OC Robotics. Visitors to the demonstration were able to take part and interact with the lead researchers. They also found out more about the underlying technologies and how robots can make us in future “super human”. The demonstration also coincided with the first “EPSRC UK Robotics Week” showcasing academic talent.

To address the technology challenge of Robotics and Autonomous Systems (RAS), researchers at UCL propose to establish several state-of-the-art experimental platforms to develop technologies for manipulating and inspecting objects remotely. 
The research at the UCL Touchlab and the Virtual Reality Laboratory bridges the gap between Robotics and Autonomous Systems and a range of real world applications. Recently funded by an EPSRC capital grant, the primary research theme is in robotic telemanipulation of objects at multiple scales – enabling exploration, manipulation, and assembly tasks in new worlds beyond human capabilities.
From molecules to the aircraft scale, technologies for telemanipulation will find applications in many fields, including healthcare, synthetic biology, advanced manufacturing, and beyond. At one end of the spectrum, the group works on devices that allow operators to manipulate micro-and nano-scale objects as if they were holding and touching them in their hands, a technology which will be useful for a broad range of applications ranging from material science to microbiology and nanomedicine. For human scale interactions, the team develops robotic healthcare tools, allowing medical interventions such as endoscopy, laparoscopy and ultrasound scans to be carried out remotely. On larger scales suitable for heavy industry and civil engineering, the team investigates new ways of controlling large robotic arms such as those needed for the remote inspection of difficult and hard to reach.

  • Collaboration: Stuart Robson, UCL Computer Science: Mandayam Srinivasan, Vijay Pawar, Dan Stoyano, Anthony Steed, Bob Sheil 

BIM and big point cloud data

Building Information Modelling (BIM) has been gaining momentum recently in the UK construction industry, especially with impending government legislation that encourages the use of this process on their contracts. The Civil, Environmental and Geomatic Engineering Department (CEGE) has seen the industrial consensus forming around BIM and felt it important to perform multi-disciplinary research in this field.

Point cloud data plays a significant role in various geospatial applications as it conveys plentiful information which can be used for different types of analysis. Semantic analysis, which is an important one of them, aims to label points as different categories. In machine learning, the problem is called classification. In addition, processing point data is becoming more and more challenging due to the growing data volume. We address point data classification in a big data context. The popular cluster computing framework Apache Spark is used through the experiments and the promising results suggests a great potential of Apache Spark for large-scale point data processing.

This work outlines a new approach for the integration of 3D Building Information Modelling and the 3D Geographic Information System (GIS) to provide semantically rich models, and to get the benefits from both systems to help document and analyse cultural heritage sites. 

Our proposed framework is based on the Jeddah Historical Building Information Modelling process (JHBIM). This JHBIM consists of a Hijazi Architectural Objects Library (HAOL) that supports higher level of details (LoD) while decreasing the time of modelling. The Hijazi Architectural Objects Library has been modelled based on the Islamic historical manuscripts and Hijazi architectural pattern books. Moreover, the HAOL is implemented using BIM software called Autodesk Revit. However, it is known that this BIM environment still has some limitations with the non-standard architectural objects. Hence, we propose to integrate the developed 3D JHBIM with 3D GIS for more advanced analysis. To do so, the JHBIM database is exported and semantically enriched with non-architectural information that is necessary for restoration and preservation of historical monuments. After that, this database is integrated with the 3D Model in the 3D GIS solution. We’ll illustrate our proposed framework by applying it to a Historical Building called Nasif Historical House in Jeddah. First of all, this building is scanned by the use of a Terrestrial Laser Scanner (TLS) and Close Range Photogrammetry. Then, the 3D JHBIM based on the HOAL is designed on Revit Platform.

Finally, this model is integrated to a 3D GIS solution through Autodesk InfraWorks. The shown analysis presented in this research highlights the importance of such integration especially for operational decisions and sharing the historical knowledge about Jeddah Historical City. Furthermore, one of the historical buildings in Old Jeddah, Nasif Historical House, was chosen as a test case for the project. 

Developments in LiDAR technology over the past decades have made LiDAR to become a mature and widely accepted source of geospatial information. This in turn has led to an enormous growth in data volume. 

The central idea for a file-centric storage of LiDAR point clouds is the observation that large collections of LiDAR data are typically delivered as large collections of files, rather than single files of terabyte size. This split of the dataset, commonly referred to as tiling, was usually done to accommodate a specific processing pipeline. It makes therefore sense to preserve this split. A document oriented NoSQL database can easily emulate this data partitioning, by representing each tile (file) in a separate document. The document stores the metadata of the tile. The actual files are stored in a distributed file system emulated by the NoSQL database. We demonstrate the use of MongoDB a highly scalable document oriented NoSQL database for storing large LiDAR files. MongoDB like any NoSQL database allows for queries on the attributes of the document. As a specialty MongoDB also allows spatial queries. Hence we can perform spatial queries on the bounding boxes of the LiDAR tiles.

Inserting and retrieving files on a cloud-based database is compared to native file system and cloud storage transfer speed.

  • Boehm, J., & Liu, K. (2015). NOSQL FOR STORAGE AND RETRIEVAL OF LARGE LIDAR DATA COLLECTIONS. ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XL-3/W3, 577-582. doi:10.5194/isprsarchives-XL-3-W3-577-2015
  • Collaboration: Kun Liu

respect to their repeatability and accuracy. It explores currently available sensors based on the 3D sensing technology developed by PrimeSense and introduced to the market in the form of the Microsoft Kinect. 

Accuracy and repeatability can be crucial criteria for the use of these sensors outside their intended use for home entertainment. The test strategies for the study are motivated by the VDI/VDE 2634 guideline. At the core of the work is the investigation of several units of the Asus Xtion Pro and the PrimeSense Developer Kit and a comparison of their performance. Altogether eighteen sensor units were tested. The results of the proposed test scenario for the sensor units show excellent repeatability at a few millimetres. However, absolute accuracy is worse and can be up to a few centimetres. Sensor performance varies greatly both for sensors of the same manufacturer and in-between manufacturers.

Point cloud segmentation is a fundamental problem in point processing. Segmenting a point cloud fully automatically is very challenging due to the property of point cloud as well as different requirements of distinct users. 
An interactive segmentation method for point clouds is proposed. Only two strokes need to be drawn intuitively to indicate the target object and the background respectively. The draw strokes are sparse and don’t necessarily cover the whole object. Given the strokes, a weighted graph is built and the segmentation is formulated as a minimization problem. The problem is solved efficiently by using the Max Flow Min Cut algorithm. In the experiments, the mobile mapping data of a city area is utilized. The resulting segmentations demonstrate the efficiency of the method that can be potentially applied for general point clouds.

  • Liu, K., & Boehm, J. (2014). A new framework for interactive segmentation of point clouds. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives, 40 (5), 357-362. International Society for Photogrammetry and Remote Sensing. doi:10.5194/isprsarchives-XL-5-357-2014
  • Collaboration: Kun Liu

A combination of faster, cheaper and more accurate hardware, more sophisticated software, and greater industry acceptance have all laid the foundations for an increased desire for accurate 3D parametric models of buildings. 
Pointclouds are the data source of choice currently with static terrestrial laser scanning the predominant tool for large, dense volume measurement. The current importance of pointclouds as the primary source of real world representation is endorsed by CAD software vendor acquisitions of pointcloud engines in 2011. Both the capture and modelling of indoor environments require great effort in time by the operator (and therefore cost). Automation is seen as a way to aid this by reducing the workload of the user and some commercial packages have appeared that provide automation to some degree. In the data capture phase, advances in indoor mobile mapping systems are speeding up the process, albeit currently with a reduction in accuracy. As a result this research presents freely accessible pointcloud datasets of two typical areas of a building captured with different capture methods and each with an accurate wholly manually created model. These datasets are provided as a benchmark for the research community to gauge the performance and improvements of various techniques for geometry extraction. With this in mind, non-proprietary, interoperable formats are provided such as E57 for the scans and IFC for the reference model. The datasets can be found at: http://indoor-bench.github.io/indoor-bench

The process of capture and modelling of buildings has gained increased focus in recent years with the rise of Building Information Modelling (BIM). At the heart of BIM is a process change for the construction and facilities management industries whereby a BIM aids more collaborative working through better information exchange, and as a part of the process Geomatic/Land Surveyors are not immune from the changes. Terrestrial laser scanning has been proscribed as the preferred method for rapidly capturing buildings for BIM geometry. This is a process change from a traditional measured building survey just with a total station and is aided by the increasing acceptance of point cloud data being integrated with parametric building models in BIM tools such as Autodesk Revit or Bentley Architecture. Pilot projects carried out previously by the authors to investigate the geometry capture and modelling of BIM confirmed the view of others that the process of data capture with static laser scan setups is slow and very involved requiring at least two people for efficiency. Indoor Mobile Mapping Systems (IMMS) present a possible solution to these issues especially in time saved.

Chadwick GreenBIM is an initiative to establish a digital 3D as-built model of UCL’s Chadwick Building including embedded environmental data to gain a greater understanding of its operation with the aim of developing concepts for retrofit and sustainable use. This paper investigates the above by reviewing the state of the art in reality capture to generate a parametric model using laser scanning. It also considers the challenges of managing information for simulation including complex ‘big data’ types (e.g. point clouds). The research in this paper was carried out through an empirical approach to create a data-rich Building Information Model (BIM) under real world conditions. Real world measurements of environmental conditions were used to validate the effectiveness of simulations using commercial off-the-shelf (COTS) software packages. In addition, the practicalities of the integration of such data inside the BIM were investigated. Data capture using the methods described in this paper requires extraordinary effort and is expensive in both time and cost. However, advances in BIM enabling technologies such as Indoor Mobile Mapping (IMM) seek improved savings in data collection. Technological advances will further improve the integration of data capture and extraction of information used by simulation tools. Once complete it is envisaged that the GreenBIM model would be disseminated among a broad community of researchers to investigate various perspectives in a collaborative manner.

Metrology, Large Volume Metrology and Digital Manufacturing

EU Project LUMINAR (Large volume Unified Metrology for Industry, Novel Applications & Research) is being undertaken to tackle several fundamental issues affecting user of Large Volume Metrology in industrial environments.  The objective of the EU-funded project is to understand and reduce environmental variations in large volume metrology (LVM). The specific task of the UCL team is to develop enhanced photogrammetric techniques to meet this.

This project will investigate and develop novel and interlinked measurement-enabled technologies for realising the next generation of factories for the "Assembly, Integration and Test" (AIT) of high value products. The vision is for the widespread adoption and interlinked deployment of novel, measurement-based techniques in factories, to provide machines and parts with aspects of temporal, spatial and dimensional self-awareness, enabling superior machine control and parts verification. The title "Light Controlled Factory" reflects the enabling role of optical metrology in future factories. A grant of £2.5million comes from the Engineering and Physical Sciences Research Council (EPSRC) and the research team, Bath University leading the project and the 3DIMPact team at UCL, have secured match-funding in excess of £1.25 million from ten industrial partners including Airbus, Astrium Satellites, Rolls-Royce, Renishaw and the National Physical Laboratory.

UCL Enterprise Knowledge Exchange Grant (Discovery to Use program) of £50,000 including match-funding from two commercial sponsors for the development of a 3DIMPact Online academy and Knowledge database about Metrology. Collaboration of UCL CEGE with Publisher Geomares (NL) and SENAI (Brazil).

This developing metrology portal aims to attract a world-wide audience and during the 12-month proof-of-concept phase must increasingly expand its public presence.

This project aims to provide a procedure for improving automated 3D reconstruction methods via vision metrology. The 3D reconstruction problem is generally addressed using two different approaches. On the one hand, vision metrology (VM) systems try to accurately derive 3D coordinates of few sparse object points for industrial measurement and inspection applications; on the other, recent dense image matching (DIM) algorithms are designed to produce dense point clouds for surface representations and analyses. This project strives to demonstrate a step towards narrowing the gap between traditional VM and DIM approaches. Geometric analyses and accuracy evaluations are performed on the output of the matching from photogrammetric 3D reconstructions (i.e. the point clouds) by adopting a metrological approach based on the use of known geometric shapes and quality parameters derived from VDI/VDE guidelines. Tests are carried out by imaging the calibrated Portable Metric Test Object, designed and built at University College London (UCL), UK. It allows assessment of the performance of the image orientation and matching procedures within a typical industrial scenario, characterised by poor texture and known 3D/2D shapes.

A DTI project to help end users understand the technology behind the techniques of large scale (large volume) metrology. A partnership between UCL, the UK's National Physical Lab (NPL) and Leica Geosystems.

Digital heritage: sites, museums and artefacts

The aim is to bring together museum conservators, educators, scientists, curators and other professionals from the higher education and museum sectors, who are interested in 3D imaging technologies and its implications across the heritage sector through workshops, practical sessions and conferences. Staff and researchers from UCL 3DIMPact are bridging the gaps between geomatics and traceable scientific surveying and cultural heritage. 3DIMPact is part of the EU COSCH - Colour and Space in Cultural Heritage, Working Group II: Spatial Object Documentation. UCL 3DIMPact / UCL CEGE is also part of the EPSRC Centre of Doctoral Training in Science and Engineering in Arts, Heritage and Archaeology (SEAHA).

Conservation of the painted ceiling at the historic Hampton Court Palace requires the condition of the ceiling to be monitored for signs of formation of cracks and flakes in the surface. Miniature monochrome cameras and LED lights were mounted on a horizontal bar, and raised by a telescopic mast to 60 cm below the ceiling. Images from the cameras were captured simultaneously by purpose developed software in a laptop computer at floor level. A series of image pairs was acquired at each location, rotating the mast between each pair. Images were enhanced in local contrast by a modified Wallis filter. From the image set a dense and accurate point cloud was generated, using a photogrammetric bundle adjustment procedure, from which 3D surface details could be visualised.

Curator David Rooney gives a guided tour of this 3D point-cloud model of the now decommissioned Shipping Galleries at the Science Museum, London. 

The Shipping Galleries at the Science Museum were 3D scanned before their decommissioning in 2012. A total of 276 scans were taken of the space and its exhibits to create a digital model of over 2 billion precisely measured points. This digital replica has been used to create a virtual flythrough of the gallery spaces with curator David Rooney providing detailed narration about the key exhibits and artefacts. The next stage of work is currently underway to launch an online, navigable version of the galleries with which the public can interact.
Archiving the Shipping Galleries digitally prolongs their legacy beyond their physical installation. As a digital record of both the space and it’s exhibits the scan model has the potential to continue this exhibition and the discussion around its content online, reaching a wider digital audience.

Collaboration: John Hindmarch, Anita Soni, Melissa Terras, ScanLAB projects, The Science Museum

Fly-through the pointcloud of the archaeological site

Jersey’s La Cotte de St Brelade is a site of major archaeological importance and is research in a collaboration between the UCL's Institute of Archaeology and Jersey Heritage. In summer 2014, Hannah Corcoran, Anita Soni and Charles Thomson flew to Jersey courtesy of the Jersey Historic Trust to consult on an Engineering Surveying project for UCL Institute of Archaeology's Matt Pope. The site is becoming unstable due to storm damage and general erosion and needs both documentation and site coordinates set up for future survey work. The survey and laser scanning produced a high-resolution coloured 3D point cloud as dataset of the cave that is a detailed digital documentation of the site, can now be used for planning site interventions. The dataset can also be used for public engagement and dissemination, and has been published in the Google Cardboard app by 3DIMPact for a Virtual reality immersive experience.

The 'Courtauld Bag', a brass bag inlaid with silver and gold and manufactured in Mosul in the early 14thcentury, is a unique object recognised by specialists as one of the most important examples of Islamic metalwork in the world. A major exhibition, 'Court and Craft: a masterpiece from Northern Iraq', was created around this beautiful object, and ran at the Courtauld institute from 20 February to 18 May 2014. As part of the exhibition, UCL's 3D imaging group were commissioned to create an animation to be displayed in the gallery alongside the object. The bag was scanned with an Arius Foundation laser scanner, imaged under a PTM1 dome and finally photographed for photogrammetric reconstruction. Despite the shiny, metallic nature of the object, a detailed 3D model was created using structure-from-motion2, while a brand new technique was used for specular reconstruction from the PTM images, creating stunning photo-realistic renderings of small details of the bag. These renderings were combined to create a two and a half minute video which was shown in the exhibition. Research is ongoing, the juxtaposition of rendered 'cgi' video and the real object affording a unique opportunity to examine and evaluate the use of modern technology and imaging techniques in a traditional exhibition environment. A prominent artist and senior research fellow at the University of the Arts, London, Jananne Al-Ani, observed the imaging, again affording a unique opportunity to explore the intersection of three disparate disciplines, art, technology and cultural heritage. Our research now focusses on use and usage of the model, as we investigate the potential of using these techniques within the cultural and heritage sector. Papers about this project present both the building and the user testing of the model, highlighting best practice and public engagement aspects of using 3D within museums and galleries.

Digital imaging technology can now produce detailed and trustworthy surrogates of historical documents. Leveraged by technological improvements in imaging and image processing, humanities scholars have been able to image, analyse, and recover more information from historical documents than was previously possible. Multispectral imaging has been utilised to examine the characteristics of documents by providing additional information about their physical properties and condition. However, current digitisation efforts have concentrated on recording documents in their current state. In this project we used multispectral imaging to record the process of macroscopic document degradation by imaging a parchment document before and after a series of degradation processes. This project is part of a larger effort to investigate the methodologies for acquiring and processing multispectral images of damaged cultural heritage documents.

The British Postal Museum & Archive [BPMA] has a comprehensive record of stamp production in the UK from the introduction of the world’s first stamp, the Penny Black. BPMA holds approximately 1300 dies and rollers from the stamp printing process and 280 printing plates. These objects, made of metal, provide a fuller context for the cycle of stamp production. The objective for this project is to provide access to these very valuable objects. The aim is to capture these finely engraved printing plates, rollers and dies as 3D digital objects. These objects pose a challenge for 3D digitization since they are manufactured mainly from steel with high reflectance, making them difficult to measure with optical surface technologies. A public outreach event will be qualitative evaluation and feedback from the public and researchers through user testing of a public delivery with selected objects from the feasibility study. The 15 Share Academy projects 2013-2014 aim to build sustainable and mutually beneficial relationships between the higher education sector and specialist museums in London. 

This project has been particularly successful to deliver impact via knowledge transfer or 3D imaging technologies to the BPMA,to foster a new integration of 3D optical imaging in the digital strategy or the BPMA, and to produce outputs that will be featured at the new Postal Museum.

3DPetrie at the UCL Petrie Museum of Egyptian Archaeology: 3D imaging research and use of new technologies in the museum. 3DPetrie is researching the viability of using high quality 3D images of museum collections to engage a range of audiences through the production of 3D models of Petrie Museum artefacts and the development of end-user digital 3D applications.