<\/p>\n\n\n\n
Large mass-engineered timber construction projects require placing hundreds of thousands of screw fixtures. A new collaboration sees a robot do the hard work.At the Murdoch University campus in Perth, a new mass-engineered timber (MET) building is being erected. The project will be\u00a0Western Australia\u2019s largest MET building, but that is not what was attracting attention to the building site one day this past May.\u00a0<\/p>\n\n\n\n
The reason for the television news cameras, the careful safety precautions, the interested labourers debating whether they could work faster than the new hire, dated from a conversation Pratik Shrestha CPEng, the Aurecon structural engineer leading the project, had a few years earlier. <\/p>\n\n\n\n
\u201cWe were throwing random ideas out and we said, \u2018Wouldn\u2019t it be cool if there were robots on site?\u2019\u201d Shrestha recalls.<\/p>\n\n\n\n
It was a \u201cblue-sky idea\u201d, he admits, but one that Aurecon wanted to see if it could turn into a reality. <\/p>\n\n\n\n
\u201cOne of the things that we do at Aurecon is we have quite good relationships with other universities,\u201d Shrestha tells create<\/em>. \u201cSo we said, well, let\u2019s do a matchmaking exercise to bring in University of Technology Sydney [UTS] \u2013 who are world leaders in robotics \u2013 partner them with Murdoch University and ourselves, and let\u2019s \u2026 see where this goes.\u201d <\/p>\n\n\n\n The end result was that trial run of a custom-designed robot built to deliver screw fixings on the construction site.\u00a0<\/p>\n\n\n\n The entire building at Murdoch University would require 200,000 to 300,000 of the screws, but on this day, the robot was tasked with affixing 50 to 100 of them as a demonstration it could do the job. <\/p>\n\n\n\n \u201cEverything ran very smoothly,\u201d Shrestha says. \u201cWe had done sufficient testing in a lab environment that when we got to site, there were no glitches.\u201d <\/p>\n\n\n\n The task is an ideal one for robots: the screws are cumbersome and numerous, and inserting them is repetitive. <\/p>\n\n\n\n \u201cThe screws that we\u2019re installing aren\u2019t just your Bunnings screws that you get for your house decking,\u201d Shrestha clarifies. \u201cThese are 300 to 400 mm long screws; they\u2019re as long as your fingertip to your elbow.\u201d The fact that the Murdoch University project used MET made it ideal to develop a robot that could take on some of the work. Shrestha compares MET to a \u201cgiant Ikea set\u201d. <\/p>\n\n\n\n \u201cIt\u2019s highly\u00a0prefabricated\u00a0and it essentially comes in a flat-pack box; can we use robotics somewhere to be able to build it?\u201d he says.\u00a0<\/p>\n\n\n\n \u201cSo then we worked with UTS to develop that idea. The robotics specialist development itself, that was all UTS, but we were highly involved in conceptualising the robot in terms of what it had to do, how it had to perform.\u201d\u00a0<\/p>\n\n\n\n One challenge the team found was that robots do not instinctively correct for unexpected forces throwing them off-balance the way a human would. <\/p>\n\n\n\n \u201cRobots don\u2019t really have that sort of gyroscopic ability to balance themselves, so one of the biggest challenges was, if the screw was slightly on an angle, it would continue to go on an angle,\u201d Shrestha says. \u201cThe other issue that we had was with the torque. With any timber construction, as you\u2019re drilling in, you need to vary the torque \u2026 because timber has varying density. It might have knots in the timber. So a robot, if you just set it to a single torque, we found that the screws kept buckling out of position.\u201d <\/p>\n\n\n\n The engineers also had to make sure the robot could navigate safely around the construction site, which involved using lidar and\u00a0building information modelling (BIM), as well as precautions such as a manual override and an automatic stop function.\u00a0<\/p>\n\n\n\n \u201cWe used our BIM model to feed into the robot to create a digital twin. So the robot, when it arrived on site, loaded up the BIM model and it knew where it was in space, and then it used the columns \u2013 the physical columns \u2013 to triangulate its location,\u201d Shrestha says. <\/p>\n\n\n\n \u201cThe robot then automatically navigates to where it needs to go, and then it starts drilling in the screw fixings to less than 5 mm accuracy.\u201d <\/p>\n\n\n\n Ensuring that navigation function worked involved thinking about the practicalities of working on an actual construction site, rather than a controlled laboratory environment. \u201cA construction site is never static,\u201d Shrestha says. <\/p>\n\n\n\n \u201cThere\u2019s people, there\u2019s mobile platforms, there\u2019s high-vis that\u2019s reflecting back on to the robot. So getting a robust enough navigation system was one of the biggest challenges early on, and with UTS, we did lots of virtual simulations using the digital twin to be able to say, okay, we can have full reliance on the navigation system.\u201d\u00a0<\/p>\n\n\n\n