In the Future, Buildings Will Be Autonomous

Machine learning is facilitating every level of smart building management, from construction to security.

By Hugues Moulin and Nicolas Sobolev, Oracle Insight

May 2018

Smart building technologies are exploding. Over the next five years, the smart building market is expected to grow by 34% annually, with a predicted total market value of US$32 billion by the year 2022. These technologies leverage an existing building’s systems information infrastructure to enable energy and operational savings through continuous, data-driven analytics and remote implementation.

However, currently, 44% of employees think their office is not smart enough, based on a study by Dell and Intel. This is not a surprise, considering 45% of all buildings stay lit even when they are closed.

But thanks to the information collected through IoT devices, companies can now take advantage of machine learning technologies to transform the way they operate their buildings. By enabling systems to take actions without any human intervention, buildings become more automated. And when these systems adapt their settings by themselves, they get even smarter and become truly autonomous.

So, how will the buildings of the future use smart technology to become autonomous? Here are a few use cases illustrating possible applications.

Autonomous construction. A building can be autonomous from the moment it is built. A measure of that has already been achieved on projects such as the 3D printed Daedalus Pavilion. An algorithm autonomously adapted the material density required for the building. At the same time, a robot fabricator with cameras connected to AI capabilities was able to judge how far its landing position for depositing material was from the design position, and thus able to correct itself. An AI feedback loop allowed it to be quicker by being more daring—it learned from its mistakes.

Intelligent energy management systems. Imagine a system that can automatically sense when a room is unoccupied or occupied and adjust heating or cooling as needed. McKinsey estimates that such systems can reduce energy use in offices by 20%, producing a potential economic impact of US$11.7 billion to US$20.5 billion per year in 2025. Intelligence can be pushed even further, moving from reactive to proactive energy management: Siemens reports the example of a solution anticipating—through employee calendars—the time, location, and number of attendees for meetings, so it can adjust temperature before attendees enter the room. Settings could even be fine-tuned throughout the meeting to compensate for the group’s body heat. Heating systems can also be adjusted based on weather forecasts or based on anticipated periods when the sun will be shining directly in the windows. And the time required to reach the target temperature can be reviewed autonomously based on past data, so the heating system can start exactly when needed.

Imagine a system that can automatically sense when a room is unoccupied or occupied and adjust heating or cooling as needed.”

Predictive maintenance. In the future, because buildings will be connected and self-aware through smart components, inefficient maintenance schemes will be replaced by machine-learning algorithms that are far more capable of knowing when preventative maintenance is needed, based on a growing bank of performance data from sensors. Predictive maintenance is typically considered for large machinery—such as HVAC systems—but it can also be applied to other use cases. In Deloitte’s Amsterdam headquarters, restrooms that need tidying give notice to the cleaning staff. U.S. Department of Energy (DOE) studies on predictive maintenance have shown predictive maintenance can reduce maintenance costs up to 30%, eliminate breakdowns 70–75% of the time, reduce downtime 35–45%, and increase production up to 25%. Additionally, facility managers achieve a 10x return on investment.

Smart security systems. Cameras have been used to monitor activity in office buildings for many years. With IoT technology such as pattern-recognition software, these security systems can be far more effective and less costly. Rather than having a security employee monitor the feed from cameras, an intelligent system can automatically detect anomalous patterns in the video data and immediately alert authorities of a possible intrusion. Samsung recently demonstrated how personal behavior can be understood based on video from closed-circuit TV cameras by using deep learning. Such security systems can also self-improve by analyzing past trends to predict future movements of occupants. According to McKinsey, smart security systems could reduce the cost of human observation needed by 20% to 50%, leading to a potential savings of more than US$6 billion in 2025.

These different use cases have in common the use of IoT devices connected with sensors. So, how can we avoid being overwhelmed with unexploited data, knowing that a single building could contain thousands of sensors? This is where big data platforms come into play, enabling the collection and visualization of data from all sources, with informative dashboards classifying the performance of individual components and the entire building management system.

Smart buildings have multiple systems, but in order to be autonomous, they need to be interconnected. As an example, Vinci Facilities introduced a new Oracle Internet of Things Cloud platform that enables remote monitoring of office equipment, analyzes sensor data in real time, and displays the results in a comprehensive dashboard. By analyzing the observed data, the platform automatically creates service requests, helping optimize repair time and improve workforce utilization.

Hugues Moulin

Hugues Moulin (left) and Nicolas Sobolev, Oracle Insight

Breaking silos to share data that was previously used by a limited audience is a good way to identify new optimization areas. At ISS, for instance, sensors on doors help commercial kitchens figure out how many meals to cook for lunch, helping cut back on food waste.

Building autonomy is becoming a reality thanks to machine learning. Beyond cost reduction and productivity gains, it is also a way to optimize environmental impact and to enhance employee satisfaction and security inside the workplace.

Action Items

Are You Prepared for Autonomous Capabilities?

Oracle Insight

Photography by Julian Santacruz, Unsplash