Integrating conversational AI and AR with BIM for faster and collaborative on-site construction assemblage

Project summary and details on research to integrate conversational artificial intelligence and National Language Processing in BIM systems.

Project summary

The impact of poor productivity within the construction industry is a global problem, which has hindered the contribution of the industry to economic development (Infrastructure and Projects Authority, 2017).

The rate at which the productivity of the industry is advancing lags behind other industries of the economy (Office of National Statistics, 2018). Lifting productivity growth by even 0.5% yearly, on a sustained basis over the next ten years would add about £7 billion to the UK economy (Infrastructure and Projects Authority, 2017). With the aim of boosting productivity, the construction industry must transform its methods of construction and adopt digital technologies.

The adoption of BIM has transformed the way buildings are designed and enhanced the implementation of manufacturing approaches within the construction industry, such as Design for Manufacture and Assembly (DFMA) (Eastman, 2011; Vernikos et al., 2014). However, the adoption of BIM by on-site frontline workers for assembly of manufactured building components is non-existent. This gap results in loss of the productivity gains accruable from using BIM for supporting construction operations (Fan, Skibniewski and Hung, 2014).

On-site frontline workers spend more time interfacing with BIM tools than they spend on completing the actual assembly or retrofitting tasks (Li et al., 2018). Current BIM interfaces are not practicable for on-site operations because they are too slow and can distract on-site frontline workers from carrying their activities (Construction News, 2018).

On this basis, the research will introduce advanced Natural Language Processing (NLP) and Conversational Artificial Intelligence (AI) for enabling on-site frontline workers to verbally communicate with BIM systems and to quickly get the information they need during on-site operations and assemblage.

Project cost

£1,860,000.

Funding body

EPSRC.

Duration

1 January 2019–31 December 2021.

Industrial partners

  • Costain Ltd
  • TerOpta
  • MobiBiz
  • WinVic
  • GeoGreen Power. 

Project details

The development of Conversational-BIM entails the following three specific objectives:

  1. Development of an AR-Based BIM Assembly Sequence Generator (AR-BIM)
    At present, assembly instructions are created throughout the ‘design’ and ‘manufacturing’ stages. Documents comprising these instructions are disjoint, wordy, and poorly written, and do not ensure errorless assembly. This project will develop an AR-based BIM Assembly Sequence Generator (AR-BIM) to ensure error-free assembly of simple and complex building components through simplified step-by-step visual illustrations on an AR device (Microsoft HoloLens); hence decreasing the rework, assembly cost by 70% and overall built assets costs.
  2. Development of a Conversational-AI System for BIM-enabled assembly (CAS-BIM)
    Currently, BIM-software provides traditional modes of interactions which hinder assembly operations automation. No BIM-systems support integrated Conversational-AI interfaces for on-site workers to navigate BIM systems. CAS-BIM will be developed to enable site workers interact verbally with BIM systems for instruction and feedback. If developed, this system will assist site workers to finish assembly tasks 50% faster through AR-assisted verbal instruction and feedback from BIM systems.
  3. Development of an Integrated Cloud-based Platform for Co-ordinated Assembly (BIM-Cloud)
    Most assembly operations involve multiple users and machines. A key innovation this project drives is to fuse AR-BIM and CAS-BIM in the cloud for a BIM-enabled coordinated on-site assembly. This integration will be enabled through various RESTful service using a combination of HoloLens AR Device, BIM collaboration tools and conversational-AI for building assembly coordination and sequencing.

You may also be interested in