Buildings are traditionally designed as static artefacts of a design process with a linear lifecycle that eventually ends in abandonment or demolition. We seek to reappraise this unsustainable approach of designing buildings which are incapable of adapting to the dynamic future socio-economic demands of an evolving society.
Game of Robots aims to enable buildings to dynamically reconfigure to the changing needs of people in a continuous and adaptive life cycle. It is an artificially intelligent distributed robotic system for the assembly and reconfiguration of architecture. An ecology of simple robotic agents that collaborate to distribute a series of simple parts through intelligent sequencing to construct spatial structures.
A series of increasingly sophisticated robotic prototypes were developed for several tasks including: autonomous 3D mobility over a compatible unit system which they are able to lock on to, crawl over, carry, and assemble; as well as carrying out collaborative tasks by connecting to each other.
The multi-agent robotic architectural system is capable of continuous reformation at multiple scales. At the building scale, a spatial assembly algorithm is used to generate effective assemblies. While at a smaller scale, portions of the design goal are distributed in real-time communication with the physical robots.
The Game of Robots system for the assembly and reconfiguration of architecture addresses context, distributed robotic systems, robotic material design, robotic automated construction, generation algorithms, and the application programme of the Game of Robots system.
A documented process of the Game of Robots system building a simple structure.
Different spatial tiles have different spatial functions and can create a variety of spatial configurations.
A bridge produced through the Game of Robots system spans a river channel.
Different materials create visual and functional variety on the surface of the building. In some places, tiles provide transparency, allowing views of indoor activities.
An illustration of the four iterations of the robots’ design and assembly.
The connection between the real world and the simulation is achieved using Optitrack – a precision motion capture and 3D tracking system for game design. Early robotic prototypes demonstrate locking, passing material, and basic collaboration.
Robots work together to achieve some actions which a single robot cannot perform alone.
Robots learn how to reach a target through machine learning agent training.
A demonstration of the full process of the Game of Robots system building a simple structure.
There are five types of units that make up the basic component types. These types are used together with an extra unit with four corners in order to allow the creation of flat ground.
Different types of units have different stabilities, weights, and degrees of enclosure.
A view of the lock and connection details. Multiple robots can co-operate together to achieve the unit connection setup.
Wood is used for the unit prototypes in real construction.
Robots collaborate to configure and reconfigure spatial tiles.
A summary of how a spatial assembly constraint algorithm is combined with stigmergy in a generative methodology.
A view of the generation of a bridge by Physarealm, combining stigmergy and the spatial assembly constraint algorithm. Several different possible outcomes are represented together.
The process of the generation of an inhabitable bridge is determined by the user’s choices such as site location, building type, material, and space selections.
A view of a final bridge assembly developed to span the River Seine in Paris.