in ECE - Data Science and Intelligent Systems, ECE - Imaging and Optical Science, ECE Spotlight Faculty, ECE Spotlight-Research, ENG Spotlight Faculty, ENG Spotlight-Research, ME Spotlight Faculty, ME Spotlight-Research, MSE Spotlight Faculty, MSE Spotlight Research, SE Spotlight Faculty, SE Spotlight-Research, Spotlight Faculty, Spotlight Research
BU Innovators Address New Requirements of Commercial Real Estate
by Maya Bhat & Maureen Stanton, CISE Staff
Smart building technology has been a growing trend in the commercial real estate sector to help building owners and other stakeholders automate processes, reduce costs, boost energy efficiency, and improve the comfort of tenants. In a post-covid world, its adoption is expected to increase as safety amenities top the list of concerns of tenants planning for re-entry. To ensure safe, healthy building environments, analysts such as Deloitte, recommend applying smart building technology to reduce and better manage occupant density and improve air quality while maintaining occupancy privacy and system security.
A new technology developed by BU CISE research affiliates from the College of Engineering has the potential to facilitate many of these re-entry preparations. The technology, called COSSY (Computational Occupancy Sensing System), has been in development through a $1 million Advanced Research Projects Agency-Energy (ARPA-E) grant. COSSY is designed to help commercial buildings save energy through occupancy-controlled HVAC. The scalable system allows building owners to estimate and monitor the number of occupants and transmit real-time occupancy data to the building automation systems to efficiently deliver airflow. It achieves these capabilities at a low cost and rapid payback, key considerations for building tenants looking for ways to recover from the economic impacts of COVID.
The project team is composed of a stellar group of researchers with deep expertise in the core technical areas required by the project. Professor Janusz Konrad (ECE), project PI, brings expertise in image and video processing and visual sensor networks. Combined with Professor Prakash Ishwar’s (ECE, SE) machine learning and statistical signal processing expertise, they work together to devise the occupancy estimation aspects of the system. Professor Thomas Little’s (ECE, SE) knowledge in smart technologies and computer networking is crucial in allowing COSSY components to communicate with each other in real time, as well as ensuring that strict cost requirements imposed by ARPA-E are met. Professor Michael Gevelber’s (ME, MSE, SE) expertise in optimizing commercial HVAC systems and measuring energy efficiency plays a key role in ensuring that COSSY delivers target energy savings.
“The system has demonstrated excellent performance during initial testing in university buildings, approaching stringent metrics put forth by ARPA-E,” says Konrad. “As we prepare to take the system for 3rdparty real-world testing this summer, we feel the timing for this technology could not be better as businesses return to their indoor workspaces.”
Improving Energy Efficiency While Managing Occupancy Density
A key priority to help enable safe office use as they are reoccupied after COVID is to carefully track and monitor workplace occupancy to ensure compliance with COVID-19 requirements, while at the same time ensuring security and protecting occupant privacy. COSSY achieves these capabilities through a highly scalable design featuring door and ceiling sensors, and precise occupancy estimation algorithms that can efficiently monitor different room shapes and sizes.
As an occupant enters or exits a room, the thermal door sensors detect body heat underneath. These sensors can work standalone in privacy-sensitive scenarios, but can also support the cameras in complex scenarios and jointly deliver high-accuracy counts.
To alert the system that a new occupant has entered or left the room, COSSY employs low-resolution thermal door sensors that are engaged when someone passes through a door. Designed for monitoring smaller spaces, these sensors capture a coarse heat silhouette of the human body without disclosing an occupant’s identity and can be deployed in privacy-sensitive scenarios. For large spaces, ceiling-mounted high-resolution fisheye cameras have a very wide field of view and can cover large areas. The modular nature of this system allows combinations of multiple door and ceiling sensors to be added and work together seamlessly in larger or oddly-shaped rooms.
COSSY relies on the team’s innovative algorithms that detect occupancy levels and combine the data from all door sensors and cameras. A fruitful combination of Professor Konrad’s expertise in change-detection, a key technology behind occupancy estimation, and Professor Ishwar’s expertise in machine learning resulted in the development of new algorithms for occupancy sensing. These algorithms accurately count the occupants in a room during both transient-state (when occupancy rapidly changes) and steady-state (when occupancy remains unchanged for longer periods of time) phases.
The fisheye cameras monitor a space, detecting and counting the people present. The wide field of view allows cameras to accurately assess the occupancy even in large spaces.
Fusion algorithms, the other key technology behind this system, interpret and combine the data between the cameras and door sensors in each room. They are critical in large-space, high-occupancy scenarios in which door-sensor or camera algorithms may err; fusion facilitates “error correction” by combining the two data streams.
Improving Air Quality and Ventilation
While numerous studies have demonstrated that proper ventilation and high air quality are essential for ensuring safe buildings, conventional HVAC systems lack the capability to monitor exact occupancy by individual rooms. Without room-specific occupancy data, HVAC systems are programmed to ventilate spaces assuming maximum occupancy. This leads to over-ventilation, the discomfort of occupants, and wasted dollars and energy.
Since COSSY utilizes its algorithms to assess the occupancy of people in specific rooms, ventilation and airflow can be tailored to room-specific requirements, ensuring high air quality and, simultaneously, improving comfort and reducing HVAC energy costs by 25% or more.
Protecting Occupant’s Privacy
In accordance with ARPA-E’s requirements, Konrad and his team have designed COSSY to be privacy adaptive. All data processing is performed within a network of COSSY sensors, and the captured data never leaves the COSSY system, i.e., no computations are performed in the cloud. With this design, all identifiable information about occupants is contained locally and only the counts are sent externally to a building automation system or to building managers for review. “The data never leaves COSSY,” assures Konrad.
Additionally, the thermal door sensors alone could be used in privacy-sensitive scenarios, such as bathrooms or changing rooms, as their low-resolution sensing only detects bodies of heat. This allows the same core technology to be applied in numerous ways in a building, all while maintaining the occupants’ privacy.
Strategic & Cost-Effective Space Management Tool
In addition to decreasing energy expenses, COSSY’s occupancy estimation algorithms provide data on the usage of building spaces, allowing chief operating officers to optimize their space management and recoup extraneous real-estate costs at an exceptionally low price.
The modular design dramatically minimizes installation and wiring requirements, making it a low-cost solution that will likely recoup the initial costs in just over a year. Following ARPA-E requirements, COSSY targets a $0.08/sq ft installed cost. “Even if we miss the ARPA-E cost target due to lack of existing network infrastructure, the technology can pay for itself in less than a year.” says Little.
The powerful wide view of the cameras is crucial in minimizing the cost of the equipment by limiting the number of necessary sensors. “If you used normal cameras, you would need 4-5 cameras to look at a whole space that a single fisheye camera can handle by itself,” says Konrad. However, if necessary, the scalable nature of this system allows more sensors to be added in larger or oddly-shaped rooms.
The precise people counting in real time afforded by COSSY should also prove invaluable for emergency response situations (fire, chemical hazard, active shooter, etc.) to ensure everyone in the building is accounted for. The numerous applications of this powerful technology have the potential to save money, space, and lives.
The researchers are currently making preparations to send COSSY to Michigan State University where it will undergo final beta testing. “By late spring, we expect COSSY will be a real-time running system. You will be able to turn it on and it will produce numbers of people in a room,” says Konrad. “We see this as a viable solution to help building owners as they prepare for the post COVID-19 workplace and are eager to make this technology available to the public.”