Drones Above: Meet the Flying Superintendent!

Royal treatment: Above, a bird’s-eye view of the Sacramento Kings’ new Golden 1 Center sports arena.

 

Eyes in the sky… 

What superintendent on virtually any large project hasn’t wanted that extra set of well-placed eyes, at least for a moment, to offer enhanced perspectives on the progress of the broader work that drives each day? And what a bonus if those eyes have been to engineering school, too.

Well, now that a robotic quadcopter drone dubbed The Flying Superintendent has taken to the skies, designed and outfitted with cameras by a trio of computer science and engineering professors from the University of Illinois at Urbana-Champaign (U of I), that bird’s-eye view is here. And we’re not talking about monthly aerial progress photos from a private plane. This is a new construction management tool that closely and frequently observes sites from above, tracking progress and enabling real-time checks against scheduled deadlines, particularly those involving time-sensitive, labor-intensive tasks.

Drawing from engineering disciplines in the aerospace, civil and environmental fields, Superintendent purports to be the first vision-based robotic system designed for monitoring actual construction progress deviations and, based on that data, predicting locations at risk for potential delays. Researchers already have successfully deployed Superintendent to track construction of a residence hall on the U of I campus, and it is currently in use on a pair of high-profile jobsites in both Chicago and Madison, WI. It is the brainchild of associate professors Mani Golparvar-Fard (civil & environmental engineering);  Derek Hoiem (computer science); and Timothy Bretl (aerospace).

Prime time sky

Flying Superintendent’s most prominent use to date has been on the new $535-million basketball arena, Golden 1 Center, now being built for the NBA’s Sacramento Kings. There, the tool has been in use for well over a year on the project that is nearing completion this summer for use next fall. There, Turner Construction is the general contractor, and the multinational firm has been so impressed with the tool that last year, it honored the Flying Superintendent with the annual Turner Innovation Award.

“The powerful thing about this is that it highlights issues with our schedule, grouped by their location, in 3-D,” explained Lincoln Wood, Turner’s San Francisco-based regional VDC manager, speaking earlier this year with the U of I College of Engineering. “[It] streamlines the management of our weekly work planning efforts by allowing us to visualize and mitigate potential risks… before they happen.”

Specifically, the web-based Flying Superintendent solution captures camera-generated images and videos and interfaces the imagery with 4-D construction software to quickly identify and visually transmit actual and potential performance problems to a project manager’s tablet or smart phone. To maintain original schedules or avoid projected delays, project executives can use the data to modify activities for more productive workflows or to improve the performance of short-term activities.

——————————————————- ADVERTISEMENT ———————————————————–

• Still need human superintendents? Let Hard Hat Hub help. To search, click here.

————————————————————————————————————————————————

The technology incorporates a drone that canvasses a site, usually during evenings or weekends for purposes of safety, taking thousands of photos that are then converted to 3-D images for purposes of detecting, tracking, and analyzing the work in progress, said Hoiem.

Resulting imagery is loaded on to web-based software that compares real-world conditions to those indicated on computerized architectural plans and 3-D visual production models. They specify “individual components, the time they should be put in place, those responsible for performing the work, and at what frequency,” Hoiem noted. As a result, project managers gain immediate access to updates on how work is progressing, including operations that are falling behind schedule or at-risk to do so.

Weekend reconnaissance

Monitoring typically is performed on Friday evenings, so that project managers have sufficient time to organize resulting data for weekly contractor coordination meetings that kick off the following week. “Discussions can range from safety issues to what must occur from day to day that week to projected delays,” said Hoiem. “Once you you’ve identified potential delays, you can work your way back and change course in order to avoid future problems.”

According to Hoiem the concept originated from Golparvar-Fard’s considerable knowledge of the construction industry, the result of his prior experience as an assistant superintendent with a large general contractor.

 

“He recognized deficiencies in oversight of site work that sometimes resulted in delays,” said Hoiem. Golparvar-Fard also understood the desire to reduce the time required to survey sites on foot. “He had a pretty good idea of the need for this type of technology, practically and theoretically.”

The three colleagues were able to fund development of the Flying Superintendent with a $1-million Cyber-Physical Systems (CPS) grant they received from the National Science Foundation for the project. Related work started in January 2015 and is now expected to extend through 2019.

Each of the three professors made unique contributions to the undertaking. For his part, Bretl’s research group constructed the robotic quadcopters and their operating system, which controls the autonomous data collection. Hoiem and his team have been working to improve the speed and robustness of 3-D reconstruction algorithms, given that the technology processes thousands of photos. Golparvar-Fard and his group, meantime, have created methods that can automatically site image-based 3-D models with pre-produced renderings of the anticipated progress, then color code them to denote site activity either ahead or behind schedule.

Getting to the roots

The system also allows project management teams to systematically conduct root-cause analysis on performance deviations by reviewing those locations that are at highest risk for falling behind schedule. So far in Sacramento, and elsewhere, the results speak for themselves.

“During weekly coordination meetings, teams can proactively address potential performance problems before they surface on the job site, and ultimately improve reliability in their short-term plans,” said Golparvar-Fard. “This is a major shift from today’s retroactive practices which involve discussing actual progress deviations. During these meetings, team can also review who is expected to do what work in what location, visually conduct what-if analysis on the plan by reviewing both actual site conditions in conjunction to the construction plan, and finally commit to the visual production model.”

At the daily level, he added that the same visual production model is used on commodity tablets to communicate work in progress. The software also captures resources used to perform tasks, and to help subs and site managers “to review actual productivity rates on a daily basis,” he said.

Those analytics conducted on survey-grade, 3-D visual production models “offer construction managers a transparent view into what’s happening on site each day, empowering them to improve reliability in short-term plans and eliminate problems before they happen,” Golparvar-Fard added.

“Great planning requires great data… data that can show the context of the work beyond BIM… data that is actionable”

Future endeavors among the three aim to streamline the data collection practice, with each developing and testing prototypes to autonomously collect images and videos both with drones and ground robots that do not depend on GPS for purposes of navigation. Included in the project are mechanisms that autonomously mount video cameras on a building component the intent being to identify and monitor construction resources while providing visual data analytics on construction safety and productivity.

“Great planning requires great data; real-time data on actual and potential performance problems, data that can be visually verified; data that can show the context of the work beyond building information models, and data that is actionable and enables root-cause assessment on performance deviations.” Golparvar-Fard said. “Our visual production management solution provides such data.”

Based on results so far in Sacramento, and elsewhere, it is hard to argue with that assessment.