ANNA

Artificial Neural Network Atmospheres

ANNA is an artificial intelligence for spatial mediation that ushers the next step in the building automation (r)evolution; something we like to call: Sensible Buildings. Its key advantage lies in its ability to blend and perform pattern recognition on existing environmental and building automation data streams. Specifically, it unifies the lattice topology of Artificial Neural Nets, with the geometry of architectural design and the subsequent spatial distribution of intelligent building components. Rather than applying AI to the input from a single sensor, it applies the basic principles of Self Organizing Feature Maps to optimize the whole.

The self-organizing principles allow ANNA to employ AI as mediator between building automation subsystems and occupant feedback loops, just as the human brain is connected to the body and its senses through nervous systems. Just as the human body, ANNA can mediate almost limitless operational data streams of its building automation subsystems, such as indoor climate, energy usage, lighting and user comfort. This allows actuators, such as smart windows to perform balanced performative operations as the skin to the body. ANNA therefor can be seen as a meta-system, that optimizes environmental energy use, in buildings and infrastructures, resulting in resource optimization through better connectivity and. system management,

ANNA enables buildings and urban spaces to performatively learn from one moment to the next, combining IoT and AI to unlock and mediate new sources of data to make buildings healthier, safer, and enjoyable for their occupants. Furthermore, the framework enables buildings to learn from each other through collaborative pattern recognition, allowing for connected building portfolios that sense, learn and act in unity. Therefore, ANNA is envisioned as continuous service framework.

Existential Necessity
Earth’s population is rapidly rising, reaching close to 10 billion people by 2050 (UN, 2019), whilst in more than 40 of the worlds most developed countries, more than 90% of the population will live in urban areas (UN, 2019). Buildings; the elementary units of these smart cities of the future will become the primary consumer of energy by 2025 (IEA et. al., 2019). Currently, we waste up to 50% of all energy and water used in buildings, leading governments worldwide to increasingly sharpen environmental performance requirements for the built environment. At the same time, commercial real estate is the second largest expense on income statements, giving companies a strong incentive to strive the same using technological advances.

The Rise of Automated and Smart Buildings
Unsurprisingly, buildings have become increasingly automated since the 1980s, allowing the aggregation and visualisation of static key performance indexes for real estate and facility management. This allowed for a better understanding and rating of buildings in relation to for example energy use, but did not provide actionable insights. After the turn of the millennium, these systems have become smarter; directly linking sensor data to analytical systems to create actionable data for building automation and control at the space and asset level, allowing buildings to perform basic actions without human intervention. However, even the smartest buildings of today, still lack the critical components to truly merge their smart subsystems and unlock their full performative potential.

Disconnected Performance
Todays smart buildings are mostly controlled by central digital direct control systems (DDC), that employ static programming principles using for example time schedules, setpoints, timers and alarms, to control aspects as temperature and lighting. Although these systems are usually customised for their initial intended use, they are rarely adjusted or optimized after installation. In part this is due to the use of only a limited number of sensors, that often lack inherent spatiotemporal awareness, to adapt spaces to yet unknown situations. This is amplified because sensor data is often collected in other locations and intervals compared to actuator devices, and their common control systems are incapable of handling large amounts of unstructured sensor data. Combined with the fact that different subsystems rarely communicate to each other, this results in that current ‘smart’ buildings translate only about 10% of the data they collect into performance operations.

Space for Living
Next to a lack in performance, and partly because low levels of interconnectivity, current smart buildings often disregard their most critical component; humans. Building controllers don’t care about comfort and health, as they usually don’t have inputs from humans. For example; human presence and the control of heating valves, window blinds and artificial lighting are rarely truly in sync. This human factor currently also poses a profound challenge for habitats in space. Although in the vacuum of space the optimization of performance and efficiency in value streams was an existential prerequisite from the very start, human comfort systems, such as lighting systems, lack automation. Due to perceived costs in designing spacecraft circuitry, these systems are mostly still hardwired, using manual switches for control (Clark 2017). Just as for smart buildings on earth, we need further automate and connect the space habitation systems, to create sustainable human environments.

ANNA’s Mission
In conclusion, current spatial automation systems, need better processing and (inter)communication capabilities, in order to reach their full performative potential and create lively environments for humans on earth and in space. Building on existing systems and technologies, ANNA’s Mission is to create a spatial ‘hyperbrain’ that adds these capabilities using IoT and AI as its foundations.

Specifications

Disciplines
  • Architecture
  • Interaction Design
  • Software Development
Typology
  • Digital
Team
Status
  • Ongoing
Phases
  • Technical Development  | 2017-01 - Now
  • Concept Development  | 2016-09 - 2017-01
Keywords
  • Artificial Intelligence
  • Big Data
  • Building Automation
  • Interaction Design