EN

Mindspaces – Inspiring Workplaces

An interactive application called Workplaces.ai has been developed by Zaha Hadid Architects’ ZH-Social Research Group as part of the Mindspaces STARTS project. The team focuses on the development of a new methodology for the comparative appraisal of the social functionality of design options, by investigating the social interaction processes to be expected in architectural environments via agent-based simulations with differentiated agent populations and autonomous decision processes. Through this process, it studies the relationship between human life process behaviour in relation to architecture to actively measure and predict social behaviour in the spaces we design. The research focuses on designing workspaces with high social performance, while expanding to investigate social interaction processes in virtual environments and events. The group aims to understand how features in built environments communicate with users and influence their collective behaviour, enabling the design of architecture with high social performance. The research involves three main components developed in parallel: 1) an agent-based computational framework for simulating and analysing the behaviour of social crowds in relation to architectural space planning. People are modelled as cognitive agents with individual internal and external state parameters, perception, and an autonomous decision-making framework to act within designed architectural environments while generating collective occupancy data; 2) a generative design model for workplace space planning, using machine learning to generate high performance layouts negotiating between a series of social and spatial performance criteria analysed through simulations; 3) an immersive experiential simulator enables users to experience and interact with designed virtual environments, while studying their behaviour in relation to them. The Workplaces.ai interactive app allows a user to enter two modes. The first enables the user to generate workplace design options by adjusting parameters which drive the model’s decisions. The second allows the user to immersively enter the virtual model, navigate and interact with it, while the social simulation runs with autonomous agents interacting within the virtual workplace environment. A series of data maps can be visualized describing the collective behaviour of the agent-based simulation in relation to the design. The program includes a menu system with options and instructions for how the user can interact with the application.

Tyson Hosmer: is an architect, researcher, and software developer working at the intersection of design, computation, AI, and robotics. His interests focus on embedding agency and artificial intelligence in nonlinear systems applied to architecture. He is an Associate Researcher at Zaha Hadid Architects, leading grant-funded research development of cognitive agent-based technologies and machine learning for generative design. He is a Lecturer at UCL Bartlett BPro in London where he directs the Living Architecture Lab (RC3), focusing its research on autonomously reconfigurable architecture, and winning the Autodesk Emerging Research Award at Acadia 2019 for his co-authored paper, Towards an Autonomous Architecture. His 13 years of experience in practice includes working with Asymptote Architecture, Kokkugia, AXI:OME, and serving as Research Director for the Cecil Balmond Studio for six years. Tyson was previously a Course Tutor with the AADRL and has been a visiting professor in several institutions internationally.

Credits

Patrik Schumacher – Chief Researcher Tyson Hosmer – Team Leader Team: Ziming He Soungmin Yu Sobitha Ravichandran Baris Erdincer