Technology

Costain and Winvic part of team exploring ‘conversational AI’ on site

Construction workers could soon benefit from voice-activated technology that beams real-time audio instructions in their earpiece, and augmented reality (AR) graphics onto their helmet visor, thanks to technology being developed at the University of the West of England (UWE Bristol).

Using artificial intelligence (AI), the system will voice and display information, removing the need for walkie-talkies or hard copies of blueprints.

The conversational AI technology is being developed by UWE Bristol’s Big Data Enterprise and Artificial Intelligence Laboratory (Big-DEAL) alongside leading construction firms, including Costain, Winvic Construction, TerOpta, Enable My Team, and Geo Green Power.

The project’s £1.86m cost is funded by UK Research and Innovation and the Engineering and Physical Sciences Research Council (EPSRC) under the Industrial Strategy Challenge Fund programme. The project started in January 2019 and it will last for three years.

Conversational AI refers to a technology that allows the user to verbally ask for and receive information in real-time from a computer. Amazon Alexa and Google Home are two examples of existing products that use this form of AI.

“Until now, conversational AI has mostly been used in labs and controlled settings,” said Professor Lukumon Oyedele, assistant vice-chancellor for digital innovation and enterprise at UWE Bristol. “Here we are bringing it into a construction environment, where workers are using their hands and need a quick and effective way to gather information.”

Typical commands could include “show me building plans” or “show me construction sequence”. The system will then supply immediate on-screen information, or instructions such as where to insert a screw in a structure. A guiding system using arrows on the head-mounted display, or a hand-held device, could also show them how to get to a specific area. 

This data can also be provided as audio information in the worker’s headphones. “One of the many challenges is to ensure that the instructions are audible and stand out, given that there is a lot of background noise on a busy construction site,” said Professor Oyedele. “We are therefore looking at technologies including noise-cancellation to allow for this.” 

As well as guiding the workers on the building site, the system can provide information for project managers, who will be able to access co-workers’ timesheets and know where they are located on site at any given time, as well as the status of various elements of the project.

“We hope that this technology will augment workers’ capabilities, to make construction more efficient. It is about improving worker’s productivity, ensuring a faster delivery process and getting it right the first time by avoiding defects,” said Professor Oyedele.

Tim Reeve, technical director at Winvic Construction, said: “It’s a real honour to be working with Professor Oyedele on his research project. AI can have relevant applications in unexpected places, and Winvic is eager to test the voice-activated headset that our data is helping to create.

“As our main focus is meeting clients’ needs – from a practical delivery point of view and also commercially – it was a natural progression for Winvic to become an early adopter of state-of-the-art BIM technology and we remain committed to digitally transforming construction.” 

Story for BIM+? Get in touch via email: [email protected]

Comments

  1. Interesting article – thanks!

    Another (potentially very useful) extension to conversational AL on construction sites is this: using the same technology, one workflow could be 1. speech-to-text, 2. machine translation into an operatives native language 3. text-to-speech. This could drastically improve the comprehension of important messages on site (i.e. health and safety) for workers who perhaps English is not their native language.

    The technology exists today – good luck with the research!

    Will (FutureDistributed.org)

Comments are closed.

Latest articles in Technology