If you ever thought flying around a construction site virtually like you do in Google Street View would be useful, now you can.
San Francisco-based OpenSpace was founded in 2017, but will start to rollout in Australia within the next 3 months off the back of a $15.9 million investment to accelerate their automated reality capture technology powered by artificial intelligence.
The technology uses 360° cameras fixed on top of worker hardhats to capture thousands of photos of the job site every day. Computer vision allows it to learn the job site, recognise objects and determine the exact location of each photo to autonomously create floor plans of the job site and align photos with the previous day's images.
The result is a virtual job site that allows project managers, consultants and clients to monitor progress remotely, walk virtually through the job site and jump back in time to compare progress or even see services through walls and floors after they’ve been covered over.
The company told This Is Construction the technology has captured over 1.5 billion square feet (140 million square metres) of projects for their customers and is providing huge benefits on the job site.
“One customer told us that they saved a project approximately one month and $1 million because we helped solve a very difficult materials installation issue. Another discovered a misplaced door that violated wheelchair access rules that would have cost $100,000 to move.”
Quality control is just one benefit, with the company saying their customers have experienced two major improvements in day-to-day operations.
The first is full transparency between the project stakeholders, which quickly puts any disputes to bed. Secondly, better communication with owners and architects expedites decision-making, reducing delays and misunderstandings.
Travel is another key benefit of OpenSpace, providing a saving of up to 50% on travel costs by allowing project and safety managers to be virtually present on more job sites, more of the time.
Consultants are also benefiting, particularly as they often aren’t allocated enough budget to be on site as much as would be beneficial. With OpenSpace, engineers and architects can keep a closer eye on important aspects of the build as they happen for greater quality assurance and to intercept problems early, saving costly rework.
OpenSpace says demand for their platform is increasing in light of COVID-19 due to limitations on site staff.
“If key team members can’t come to the site then OpenSpace becomes a key enabler to simply keeping the ball rolling and preventing the job from just getting suspended.”
How it Works
Any worker on site mounts a 360° camera on top of their hard hat (the OpenSpace system works with any 4K 360° camera). The worker launches OpenSpace’s mobile app on their phone, provides details about the job they are working on, and then press ‘Start’ to begin the capture. The capturing process is fully passive so the worker can continue their work without operating the scan.
When the job is finished, the worker hits ‘Stop’ on the app, and then uploads the data to OpenSpace’s cloud for processing. This usually takes in the order of minutes, not hours, but is dependent on the size of the scan and the frequency of previous scans.
The more scans that are uploaded, the quicker the uploading process becomes because the AI is programmed to ‘learn the site’.
Once that process is complete, the scans can be viewed at a custom URL that can be accessed by the whole project team. The user can then look around, click through a floorplan to move around to any location or take a virtual walkthrough.
The platform isn’t simply a reality capture tool, but uses AI to interpret that information with features such as ‘Object Search’ allowing the user to search the whole job site for a particular object such as an emergency exit sign or a piece of equipment.
The BIM Compare feature allows the user to compare site images with the corresponding BIM image side-by-side.
ClearSight Progress Tracking uses AI trained to recognise items, such as wall frames and plasterboard, and allocates a % complete to the relevant location on the plans.
The platform also integrates with several design and project management tools such as Procore, PlanGrid and Autodesk’s BIM 360.
Several other companies are targeting the reality capture and autonomous progress tracking space including Disperse, Doxel, HoloBuilder, Struction Site and Scaled Robotics, some of which use robots to autonomously roam the job site capturing images.
OpenSpace says that their platform’s main advantage is its ease-of-use and ability to plug-and-play. OpenSpace can be used with any commercially-available 360° 4K camera and doesn’t rely on location-tracking or GPS to align images. Instead, its computer vision ‘locates’ itself through image recognition - meaning there is no extra integration required for location-tracking devices.
“OpenSpace can be used TODAY, out-of-the-box, on any mobile platform you choose, including robots. The fact that our system uses computer vision to automatically locate and align images means that anyone (or any thing) can walk the site, on any path, and it will be aligned and registered to the floorplan.”
OpenSpace is charged based on a percentage of project value, which the company wouldn’t reveal but they assure us it is below 0.1%. This would be somewhere under $100,000 on a $100 million project, which is seemingly money-well-spent to avoid construction errors with the potential to cost millions.
There is also a free plan available providing 360 site capture but requires manual placement of images on site plans, rather than the automated placement with paid plans.
Will it Revolutionise the Construction Site?
It’s a no-brainer that having more eyes on the site, more of the time, without restriction and costly travel time is going to have many benefits. These will stem from improved communication and dispute resolution, early identification of errors, and more efficient scheduling and coordination.
This all translates to time, cost, quality, safety and sustainability objectives.
But what will truly revolutionise the construction site is what’s next. This technology opens up a gateway to something bigger than data collection: data interpretation.
OpenSpace has already started down this track with object recognition and progress tracking. Whilst these platforms may start as reality capture, they will develop into site analysis tools that don’t just collect data, but are able to analyse it and draw conclusions to pinpoint areas for improvement and provide actionable recommendations on more effective job site operation.
These tools will harness machine learning to become recommendation engines, advising project managers on operational improvements such as the most efficient site layout, waste reduction, critical path scheduling, crane and plant movements, etc. - the possibilities are endless.
Cofounders Jeevan Kalanithi (CEO), Philip DeCamp (CTO) and Michael Fleischman (Chief Scientist) all have extremely impressive resumes, with experience ranging from PhD’s to previous company acquisitions by Twitter.
They have raised a total of $33 million since setting out in 2017 to solve what Kalanithi says is one of construction’s biggest problems: record-keeping. But judging from the extremely high capability within the founding team, don’t expect them to stop there.