Construction Co-bots Need To Be Smarter, More Adaptive (Part Two)
For robots to really change the construction industry, they need to become more than just ‘tools’ in the eyes of the people using them. Previously I wrote about two construction robotics companies that are augmenting construction workers to alleviate a serious skills shortage, but these solutions focus on providing more powerful tools on-site, rather than looking further afield at the construction industry’s potential for automation.
In this article, I will delve deeper into robotics and automation in construction to look at how true autonomy might be achieved in a complex and high-risk environment, and what impact robotics might have on workers and the industry as a whole.
Adaptability is king
As AI becomes far more capable of dealing with nuanced and high-pressure situations without human intervention, more traditional pre-programmed industrial robots are becoming less useful in comparison. “Six-axis industrial robots are largely ‘mute’,” says Erin Bradner, Director of Robotics at the Autodesk Robotics Lab, “they are blind, they have no sense of touch, and [can’t really] respond to their environment.” As sensors are embedded into robotic arms and mobile robots as well, there is an increased emphasis on “introducing adaptivity into robots, to move beyond the paradigm of pre-programmed motion,” says Bradner.
The utility of traditional industrial robots or robotic arms has been proven over the years, but with an increasingly digitized working environment and severe skills shortages-particularly in manual industries such as construction-more competent and independent automation is needed to keep up with the pace of innovation.
The limited ability of stationary robots to respond to their environment could be a huge hindrance in a scenario such as a busy construction site, leading to more work for those using the robot to allow it to operate effectively. Mobile co-bots that roam around the job site are far more responsive to their environments, but this can also be a hindrance if they aren’t smart, as Bradner explains: “Co-bots are designed to be human-safe, they’ll stop when they encounter an obstacle, but you don’t want your project stopping every time a person sneezes.” To get past this deterministic hyper-vigilance, some co-bots like those of Scaled Robotics and Dusty Robotics are introducing adaptivity and more nuanced intelligence so that, as Bradner puts it, “the co-bot will be more flexible and responsive to its environment so that humans don’t have to adapt to the co-bot.”
Finding new paths
There are a few possible ways for a co-bot to respond to its environment, either “stop whenever there is a force exceeding a certain threshold,” “teach a robot to understand what a dynamic environment feels like,” in the cloud and maintain communication in real-time, or what Bradner calls the brute force approach, “burning into chips the ability to follow every single path and allowing it to choose the right one.” Each of these approaches has its merits and drawbacks-the first is simple but can’t judge outside of the “force” parameter, the second is more nuanced but needs to communicate with the cloud, and the third takes a huge amount of processing power to give the most possible autonomy-but the question remains which method is best or which combination will suit different applications.
Ultimately, Bradner states, “for a robot to adapt to more dynamic environments, we need to train it to be able to infer what happens if something crosses its path, how to respond to that, sense it, plan and act in that context.”
Given the complexity and intensity of a live construction site, it is likely that a combination of these approaches will be needed to reach a workable solution for jobs where robots are manipulating building materials on-site.
There are many applications for robotics away from the construction site itself that might provide a quicker win for automation, stemming from the wave of digitization that has swept the industry over the last few years. Bradner points to the prevalence of incredibly granular CAD models that are used ubiquitously throughout the industry, and the possibilities that such a level of digital insight off-site could open up for robotics. “When a CAD model is handed over to a contractor or a manufacturer, typically the geometries and tolerances in the 3D model are just used as a guideline. What if we could leverage the inherent logic and accuracy of a 3D CAD model, explode the model, and then simply train a robot to reverse that explosion. You’ve just created a set of perfectly accurate assembly instructions that a robot can follow to the letter.”
Assembling the future
Developing robotics construction in this way, using automation, digital software, and bypassing manual processes and errors, sets an interesting foundation for the future of robotics in construction. Not one where robots wearing hard-hats replace workers, but one where the entire construction site itself changes to more of a process of assembly, rather than construction as we think of it today.
As automation and robotics reach deeper into an increasing number of industries, the question is whether a particular industry adopts robotics into its processes, or changes itself completely around a new way of working with robots. Facing a lack of skilled labor and a shortage of resources, automation might help the construction industry to rebuild itself from the ground up.
Originally published at https://www.forbes.com.