skip to content

EPSRC Centre for Doctoral Training in Future Infrastructure and Built Environment

 
None

Computer vision and real-time tracking applications, finite element simulations, big-data crowd-sourcing.

Theme 8: Computing Technologies in Engineering

 

Project Title

Enabling teleconstruction via robotics teleoperation

Primary Theme

Computing Technologies in Engineering

Secondary Themes

Construction Design and Technology, Asset Management

Project Summary

Unlike the manufacturing industry, the construction industry does not build most of its products at a controlled plant facility. Construction workers operate mainly outdoors, which exposes them to occasional extreme weather conditions. Some of the work can be and is gradually moved off-site, but the bespoke nature of the industry will always leave a substantial portion of the work to be carried on-site and maintain the industry’s terrible safety and productivity record. Manually operated construction equipment (loaders, dozers, etc.) have increased productivity and safety on-site to some extent; yet the industry is still far from achieving the standards of the manufacturing industry. The objective of this project is to devise, implement and test a new theory for moving the human construction worker off-site (to an office), and link his motions and senses to those of a robot on-site that will build instead of him/her.

 

 

Project Title

Robotics prefabrication

Primary Theme

Computing Technologies in Engineering

Secondary Themes

Construction Design and Technology, Asset Management

Project Summary

A survey carried out in 2007 among UK government departments, developers, consultants, main contractors and sub-contractors showed that prefabrication is viewed as inflexible for changes of design. This lack of flexibility is ranked as the most important barrier to market. The respondents argued that consultants and clients are reluctant to adopt it because of the need for an early design freeze. Companies that bet on prefabrication are rushing to develop design solutions that can accommodate flexibility. However, the same is not taking place at the fabrication level, with not very intelligent factory robots that need substantial resetting and reprogramming with every design change. This project will investigate the feasibility of building a virtual robot that will not only build a virtual structure given an early design, but will also be able to respond to design changes in real time, deconstruct parts no longer necessary, and fix mistakes. The focus will be on automatically translating a product (Building Information Model) into a process (set of building instructions). The project will deliver a prototype software implementation of the proposed solution as well as

a PhD dissertation describing the need, the state of the art, the new solution and the results.

 

 

Project Title

Digitizing tunnels

Primary Theme

Computing Technologies in Engineering

Secondary Themes

Construction Design and Technology, Asset Management

Project Summary

The lack of good quality underground infrastructure maps/models is a well-established problem in the UK. This is particularly felt in areas like London where tunnels and utilities criss-cross below each other in numerous locations. Various methods have been employed recently to map mostly new tunnels and utilities. Yet none of them has provided adequate value for money to tunnel owners and other stakeholders to address this matter effectively for both existing and new tunnels. For example, photogrammetric solutions are now able to provide a locally accurate surface map of the inside of tunnels at an accuracy competitive to laser scanners. Yet, both point clouds and photo models are raw, low-level geometry datasets far from the geometry level of a design model and even further from the upcoming Level 3 BIM standard. The objective of this project is to devise, implement and test a new method for automatically generating geometrically rich information models of existing and under construction tunnels from point cloud and image datasets. The goal is to devise a solution that can generate a highly accurate and reliable BIM file of a tunnel using the IFC-INFRA standard with negligent human input. The resulting model will be geometrically accurate and rendered with the available visual data. The project will deliver a prototype software implementation of the proposed solution as well as a PhD dissertation describing the need, the state of the art, the new solution and the results.

 

Project Title

Cloud BIM

Primary Theme

Computing Technologies in Engineering

Secondary Themes

Construction Design and Technology, Asset Management

Project Summary

Current generation software packages used for modelling existing structures are not fit for purpose, as they are unable to handle the computational, storage, security and interoperability needs of the built environment. New methods are required, that can make effective use of current and future high performance computing platforms to generate as-is models at a massive scale. The major challenge is the sheer volume of computations. For instance, Point Cloud Data (PCD) from a single industrial facility can have many billions of points. Likewise, a simple HD (2MPixels) video stream of the same facility can have tens to hundreds of billions of pixels. Processing such data is a major computational challenge; all raw data must be cleaned first through multiple and often independent processes (denoising, outlier removal, enhancement, etc.), followed by applying computationally intensive learning processes. If we then multiply that with the massive number of models needed in the future to employ effective learning through big data, the problem becomes truly substantial. This is further compounded by reasonable concerns of data security and resilience over the life-cycle of the infrastructure. The objective of this project is to devise, implement and test a new framework for migrating infrastructure models and the modelling process to a high performance computing (HPC) enabled cloud. The goal is to enable HPC efficiencies at the server-cluster level (with millions of computing cores) to undertake hitherto intractable a) fabrication-accuracy 3D reconstructions, and b) training/testing of pattern recognition models using volumes of partner-provided models. The project will deliver a prototype software implementation of the framework as well as a PhD dissertation describing the need, the state of the art, the new approach and the results.

 

Project Title

User monitoring in buildings to inform structural behaviour

Primary Theme

Computing Technologies in Engineering

Secondary Themes

Sustainability and Urbanisation, Construction Materials and Waste Minimisation

Project Summary

We poorly understand how the behaviour of building structures affects building users, including in crucial areas of health, wellbeing and productivity. These are core parameters behind structural design, yet presently are poorly understood. It has been demonstrated that imperceptible floor vibrations can have health impacts; and that office workers take more days off sick the higher in a tall building they sit. This project aims to utilise continuously synchronised structural and human monitoring to learn from the behaviour of buildings to inform appropriate serviceability levels in structural design. This has never been done before, but is of crucial importance as increasing research focus is placed on reducing the embodied energy of new building structures to reduce overall CO2 emissions from the sector, and is driven by lightweight design. The project has extensive theoretical, analytical, and experimental components.

 

Project Title

Peridynamics for structural concrete analysis

Primary Theme

Computing Technologies in Engineering

Secondary Themes

Sustainability and Urbanisation, Construction Materials and Waste Minimisation

Project Summary

Concrete is the world’s most widely used man-made material. Its use underpins the productivity of other industries. Yet the manufacture of cement is responsible for at least 5% of CO2 emissions, and research has shown that material wastage in the order of 50% is common. To reduce this wastage, we must optimise structural designs. Such lightweighting requires robust analysis methods. Finite element methods are not well suited to the analysis of brittle materials such as concrete, as they assume continuous body deformation to calculate partial differential equations. An alternative solution may lie in peridynamics, a novel meshfree analysis method. Peridynamics is based around an integral formulation, valid even if disruptions of the material continuity occur since no space derivatives are employed. Solid materials are modelled as particles held together by tiny forces, the values of which are a function of each particle’s relative position. By defining the interaction behaviour of different sets of particles, multiple materials can be modelled. This PhD will build on research at Cambridge to expand the understand of how peridynamics can be applied to reinforced (and prestressed) concrete structures, with the specific goal of using this powerful tool to define new optimised geometries that minimise material use.