On-Site Supersensing and the Future of Automation: Discussing Construction, Robotics, and Contemporary Data Capture with Aviad Almagor

Is there anything practitioners can be doing to facilitate this atmosphere of change?

First of all, it’s our responsibility to provide the confidence that the technology can indeed serve the end user. We’re doing this by investing in educating the market about the technology and its capabilities. This means creating close relationships with customers and working to really understand their needs, their pains, and then to gradually expose them to the technology. This helps us figure out what is working well, what is not, where are the technology gaps, and what we need to improve in order to make the tool really efficient. Then it becomes a journey that we are on together with the customers from an early, visionary concept to implementation, and the successful end-use of the technology.

I presume our current era also provides its own challenges?

Of course – we’re in an early phase of robotics and part of the challenges are to set the right expectations and to identify use cases where the technology can be already useful. For example, let’s take contemporary robotics. We can identify where robots can perform dull, repetitive and dangerous tasks, and direct our efforts in order to fulfil requirements and support tasks. This could be as straightforward as monitoring a site instead of sending a human out to do repetitive work, which may carry some risk. This can of course be manually operated, but the introduction of automation is an example of the low-hanging fruit we can already support with technology and then implement together. So, today, I can take a robot. I can teach that robot a task. That robot will follow their set path quite effectively using the special understandings and mechanisms we already have to identify features along their set path. But, as we all know, construction sites are very dynamic and change over time – often on a daily basis. Imagine we need to send this robot to a different environment with little warning. Here AI and semantic models and machine learning processes help us better understand the environment, which then helps the robot better understand their environment, and resist daily disruptions that would derail performance.

What technology helps the unit obtain this data, and how is the data validated?

Data capture is vital. The longer the range of the reality-capture sensors on the robot, the better. Because it will enable the robot to better understand the environment. This is especially important in areas where you have repetitive objects. For example, a large floor with repetitive columns. If you are a robot, how do you know where you’re standing? Longer sensor range helps provide this fuller picture and integration with design data also helps improve our understanding. This could include adding a 3D model into robot navigation maps which then enables them to identify alternative routes in case of unforeseen challenges. This could be as simple as a pile of material blocking what was meant to be a clear route or dealing with significant risks like rebar cages. But of course we need to go back to the users, because we don’t care about the robot itself.

What we care about is the output that a unit can deliver and how that ultimately works to support decision makers. This element is highly task dependent and for that we use different sensors. We can use RGV cameras to monitor and communicate easily with all stakeholders in the project based on daily pictures that the robot will take. We can use highly accurate reality capture devices like laser scanners with millimeter accuracy and a range of 100 meters. These tools can be used engineers or contractors to better understand the accuracy of the installation and monitor the progress of the work.

Did anything about the pandemic challenge your ideas and preconceptions about the industry?

Actually we were surprised by the demand for robotics capabilities during the pandemic. The reason for this was that, with robots physically on site, you can safely carry out site visits. You can conduct remote monitoring and provide stakeholders with an ability to remotely immerse themselves on the site, and communicate with the teams while remaining in a safe environment. So, in this respect, the pandemic actually was an interesting kind of motivator to push robotics and automated processes.

Do you feel this is a pattern for the sector? Or do believe there are other trends that we should expect robots and AI to tackle? 

I wish I knew the full answer [laughs]. But what we absolutely do need to do is look at this in the light of other trends. These don’t have to be technological trends, but certainly business trends. We’re recently seeing a shift from on-site construction to prefabrication. I believe we’re going to start to see construction sites turning into assembly sites, which will in turn will help robotics to be more efficient. This means more automated processes, autonomous machines working on side and on the side, much better predictive and prescriptive processes to support a more efficient environment.

Finally, can you give us a taste of what you’re going to cover for WDBE 2021?

Definitely – it will be an opportunity to learn more about the unique partnership between Trimble and Boston Dynamics and how we bring to the domain knowledge from both sides in order to really help transform the construction industry. I will not say more other than this is going to be an exciting keynote. So, I encourage you all to join us.

Aviad Almagor will be speaking at WDBE 2021 on September 29, 2021. You can book tickets from here or learn more about the other events and the regularly updated agenda.