unit-code
Reconfigurable Autonomous Demiurge (RAD) proposes a living architecture with the ability to adapt to changing circumstances, revolutionising the design process and life cycle of the built environment.
RAD is a modular architecture consisting of flexible components with reversible joints allowing their assembly, disassembly, and reconfiguration by distributed robotic agents. To organise the phases of design at multiple scales, an architectural language was conceived where components and materials are organised following the logic of real language structure.
Robotic agents were designed to assemble components while using them for navigation, saving on human labour and reducing waste. The robotic system was evolved through a series of prototypes while incorporating AI through deep reinforcement learning to train robotic agents in a simulator to learn adaptive and collaborative policies for reconfiguring spatial assemblies.
The platform in formats of App and Web allows users to interact directly with the RAD system. It allows people to participate in the process of design, customise their space, and create their own furniture. This enables anyone to be the designer of their living environment.
RAD proposes a new way of perceiving, creating, and occupying buildings, where computational tools, robots, and humans coexist and co-evolve enabling the built environment to be an adaptive ecosystem.
Reconfigurable autonomous demiurge.
RAD project lays on the idea of living architecture, where the life cycle of a building is continuous and not determined.
With the employment of robots in RAD system, an adaptive architecture emerges.
According to the scale and the stage of the process of the design, materials and parts used in it are organised following the logic of the real language structure.
A cubic node and a linear timber component constitute the base of the language.
The cubic node functions as a joint to assemble components. The components slide inside the cubic node and through a locking mechanism they are locked in their final position.
Materials can be applied on the rails existing on the components to customise and give different qualities in the space.
Combination of several components to form bigger aggregations - the words. Several of these aggregations are combined to form a sentence.
A series of robot agents are developed to assemble and disassemble. The latest robots are in modular units and collaborate in each locomotion movement.
Robotic agents in charge of carrying, placing, and assembling the alphabets.
Through machine learning, agents were trained to navigate around the structure and reach their target.
Robots cooperate to achieve their actions.
Suburban mode in the platform where users can edit the space and the spatial algorithm solves accordingly. Different environments can be selected to build the space in.
The user can defines the spaces they require; a force based graph generates a relation-graph layout within set boundaries. The user can then drag and move these nodes to their desired position and hit 'generate'.
Our space making algorithm can process voids, solve and create cores for different zones and connect them by generating corridor spaces. The program also analyses the structure and places support structures where necessary.
A builder engine is developed where different parts with varying connecting rules can be introduced into the engine and the user can build or generate unique and self made aggregations based on how they connect with each other.
Our complex space generation algorithm uses Bit masking values to dynamically solve the structure in real time to make a thorough and valid space.