01

Case Study Illustration

Case Study Overview

Case Study Overview

Case Study Illustration

Case Study Illustration

Diffusive Habitats is a comprehensive architectural system that utilises reinforcement learning, AI, and robotic assembly. The case study illustrates the holistic process and provides situated scenarios to show the potential of the system.

02

Algorithmic Research

Algorithmic Research Overview

Algorithmic Research Overview

Spatial Planner Algorithm

Spatial Planner Algorithm

The Spatial Planner Algorithm starts off Diffusive Habitat’s algorithmic process. Within it, individual spaces grow and negotiate to optimise space under a series of situated conditions.

Reinforcement Learning - Agent Behavior Development

Reinforcement Learning - Agent Behavior Development

To encourage collaboration, the agents are trained to carry out specific tasks using a weighted reward system.

Spatial Assembly Algorithm

Spatial Assembly Algorithm

Spatial assembly physically renders the result of spatial planner. Utilising various techniques found in procedural generation algorithms, spatial assembly manipulates the curated tile set to solve for the constraints given by the spatial planner.

03

Material Research

Material Research Overview

Material Research Overview

Material System Details

Material System Details

Various versions of the primary units have been designed to solve specific conditions found within the system. These custom units include: MEP services, solar panels, surface charging, multi-material, and adaptable storefront units.

Material Prototyping

Material Prototyping

Material prototyping has played an instrumental role in the design and robotic testing process. Frequent alterations and modifications have considered many factors, for example the weight, stability, and size.

04

Mechatronic Research

Mechatronic Research Overview

Mechatronic Research Overview

Mechatronic Design

Mechatronic Design

Diffusive Habitats’ mechatronic design was conceived by analysing, merging, challenging, and furthering state-of-the-art robotic morphologies. At its core, the design pursues simplicity and collaboration, while minimising motor forces and torques.

Mechatronic Testing

Mechatronic Testing

Systematic experimentation (gauging the capabilities and limitations of the robotic prototypes) proved fundamental for prototype evolution, revealing the vast flexibility and unforeseen capabilities of collaborative behaviours.

3D Tracking

3D Tracking

Iterative studies on robotic control led to reactive 3D-tracking with Unity and Optitrack. With it, Diffusive Habitats is able to motion-capture and coordinate multiple robots in real-time, while receiving precise positioning feedback data.

05

Intelligence Research

Intelligence Research Overview

Intelligence Research Overview

Reinforcement Learning Arena

Reinforcement Learning Arena

The project utilises a game-based research framework to develop its robotic intelligence. This arena illustrates the process by which AI is trained to compete against humans and hardcoded algorithms within that framework.

Applied Intelligence | Skylight Simulation

Applied Intelligence | Skylight Simulation

The framework’s best performing strategies can be applied to larger scale reconfigurations. For instance, this 6-robot simulation features AI-based pass behaviors, pathfinding algorithm sequences, and human-derived multi-agent “turn” coordination.

Applied Intelligence | Reconfiguration Studies

Applied Intelligence | Reconfiguration Studies

Likewise, this intelligence can be applied to the project’s physical prototypes and physical studies.

Reconfiguration Studies Catalog

Given the nature and flexibility of Diffusive Habitats' system, material can be reconfigured into an array of shapes to serve diverse functions and account for changing needs.

Reconfiguration Studies - Arch

Physical reconfiguration studies gauge constraints and inform the computational models. For instance, this 50-unit arch could not be assembled as initially calculated, revealing much about the sequencing and reinforcement of cantilevering structures.

06

Architectural Experience

Architectural Experience Overview

Architectural Experience Overview

Diffusive Habitats engages in improvisational, action-centric, and situated activity. The temporal nature of the structure allows for various readings across multiple scales of time.

Individual

Individual

Any given configuration can be interpreted as a result of negotiations between multiple agents. In the case of primary living spaces, the occupants have a direct connection with the architecture's configuration.

Collective

Collective

Larger configurations decided by the community reflect a multitude of agents and time. The changes are caused by and effect inhabitants collectively; changes on this scale might be described as seasonal or regular.

Urban

Urban

The urban scale is still fundamentally shaped in an ad hoc fashion. However, at this scale, the result conveys a much larger succession of moments in time.

Beyond Architecture

Beyond Architecture

The primary actions of the robots revolve around reconfiguring material, robotic choreography and dance studies, as well as interactions between human and non-human agents.

Share on Facebook, LinkedIn or Twitter