technological innovations

Technological innovations in condominiums

The future of drones, robotics and software in building envelope inspections
Thursday, September 29, 2016
By Filip Sobotka

Technology is one of those things that never stops moving, and sometimes it can be daunting to keep up with all the latest advancements out there. If leveraged correctly, though, it can provide great amounts of value to those that take the jump and never look back.

Everyone loves technology, but it takes a little bit of foresight and a change in perspective to appreciate the potential value. Take the example of the soon-to-be self-driving car. On the one hand, it might sound silly to have a car controlled by a network of computer programs. On the other hand, it might eventually seem insane in retrospect that 16-year-olds were given control of a 4,000-pound machine, capable of hitting speeds of more than 100 kilometres per hour.

So, what equivalent technological innovations are happening in the world of condos today?

Data collection

Technology is being leveraged to conduct dangerous jobs such as building envelope inspections. Consider the case of the engineer on a swing stage taping his camera to a broom stick to get a view around the corner of the building.

The advancement of drones and robotics allows for this collection of information to be done faster and without the hazards associated with dangling humans from heights. Instead of collecting data from a small portion of the exterior and extrapolating the results, it’s now possible to capture information from the entire exterior.

When it comes to the health of a building, more information is always a good thing, and a drone can easily fly around a building and collect data across the envelope.

Computer vision and machine learning

Now, with all these images of a building to review, technology can be of further assistance. Computer programs are being developed to perform more complex tasks traditionally reserved for humans.

Two-dimensional images captured from a camera mounted on a drone or robot can be run through image processing and analysis for contrast enhancement, edge detection or noise removal. It’s like zooming in on an image using a microscope and grouping together or clustering similar groups of pixels as a means to detect features. This method has been successfully applied already to identify defects and assess the condition of concrete and asphalt in civil infrastructure.

Machine vision takes this a step further by using various techniques to automate imaging-based inspections. Being more complex, this often requires the help of humans initially, in a process called supervised learning, where humans provide a filter for the image processing system.

An inspector or engineer can look at an image and can identify it as a window, or as a crack in a brick. Then, after the computer has learned this feature — often after thousands of instances — the computer can do it autonomously. This becomes valuable because now the system can save human inspectors time spent reviewing images by filtering them for issues. This technology is still in its early stages, but it’s already being applied in the medical industry to brain and eye imaging, where algorithms can predict diagnoses by piggybacking inputs from specialists based on previous similar and learned images.

The building envelope presents a different set of challenges to the use of machine vision due to the exterior environment. Not being in a controlled setting means the system needs more images and time in order to learn and become more sophisticated.

The real fun starts when predictive models can be applied using artificial intelligence (AI), layering more data, such as weathering patterns and free-thaw cycles, onto the model. Frequent and repeatable scans completed using robotic and drone technology can detect the changes over time, with what is known as time-lapse or fourth-dimension photography.

This learning model allows for humans to input what their previous knowledge suggests should occur, and then get real-time feedback on what actually happened. This information is then returned to the feedback loop and the process starts all over again, bettering everyone’s knowledge each time.

What’s coming next?

The technology train never stops and the systems will only continue to improve, reducing their error levels to match human performance and even surpass them at some point. The other major trend is the development of sensor packages on these robots and drones — laser measurement, depth sense sensors, and sonar capabilities, just to name a few.

The challenge is translating this insight into valuable information. Each step forward in sensor development requires a step back to integrate the new information and capability.

The smart folks at Stanford have combined the climbing robot with drone capabilities with their SCAMP (The Stanford Climbing and Aerial Maneuvering Platform) project. Using gecko adhesives to climb along a wall, the drone bug also has the capabilities to disengage from the wall and fly to safety should slippage occur.

Now it’s possible to begin looking at what sorts of knowledge can be gained from using this new ability. Perhaps pressure sensors could be used to detect loose components that pose a fall hazard? The possibilities are really only limited by the things humans can dream up.

As discussed, appreciating the value of leveraging technology is all about perspective. Some may think that all this stuff is outrageous and that it’s not feasible to ever eliminate the need to dangle humans off the sides of buildings. Others may just be look up into the sky one day, see a drone flying around a building and think to themselves, “How did we ever scan and review these buildings without these things around?”

Filip Sobotka is co-founder and CEO of FTD Highrise Inspection Inc.

Leave a Reply

Your email address will not be published. Required fields are marked *

In our efforts to deter spam comments, please type in the missing part of this simple calculation: *Time limit exceeded. Please complete the captcha once again.