Abstract
Architects and building owners often lack a systematic way to understand how their buildings are used, making it difficult to improve designs. To address this, the Sensing Behaviour framework provides methodologies and tools to gain insights into the behaviour of building users. The framework connects pre- and post-creation loops to learn from past examples and inform future designs. Sensing Behavior uses object and relation detection algorithms on video streams to understand occupant behaviour. The framework couples detected objects and persons and their interactions as nodes and edges in a time-based graph representation. This paper describes the technical development of Sensing Behavior, related workflows, and protocols, and its implementation in a real-world use case as proof of concept. The framework’s capability to record and analyze occupancy behaviour has the potential to inform future architecture and create sensor systems for better building design.
Original language | English |
---|---|
Title of host publication | 2023 Annual Modeling and Simulation Conference (ANNSIM) |
Number of pages | 13 |
Publication date | 26 Jun 2023 |
Pages | 584-596 |
ISBN (Print) | 979-8-3503-1457-1 |
ISBN (Electronic) | 978-1-71-387328-0 |
DOIs | |
Publication status | Published - 26 Jun 2023 |
Event | SimAUD 2023 - Hamilton, Canada Duration: 23 Jun 2023 → 26 Jun 2023 |
Conference
Conference | SimAUD 2023 |
---|---|
Country/Territory | Canada |
City | Hamilton |
Period | 23/06/2023 → 26/06/2023 |
Keywords
- Behaviour
- Occupancy Analysis
- Computer Vision
- Machine Learning
- Architectural design
- Object Detection
Artistic research
- No