top of page

A Generative Approach to Robotic Fabrication

Adaptive Growth Method vs Plan Execution Method


This project explored computational methods for adaptive growth seen in some human design processes, such as development of spontaneous settlements, through a relatively simple yet explicit example in the context of robotic fabrication. The proposed experiment uses an industrial robot arm to produce structures by stacking unit bricks without comprehensive hard-coded instructions for construction (“blueprints”) from the outset. Using programming in Java and RAPID, the project explored an application possibility to obtain a more flexible and open-ended way to send instructions to the manipulator. The program of the experiment can produce structures that satisfy a certain characteristic – a simple rule of physics and a footprint of the site – while maintaining some level of morphological variations using a stochastic selection process. Every result of the program differs due to the stochastic nature of the program. However, all results satisfy the same initial footprint condition defined by a user and the premise that the robotic arm can build a well-balanced structure by stacking unit bricks on the fly on its own. This operation can be done without hard-coded target positions of all bricks from the outset of the process. The system can find its next position as it proceeds without having a fixed blueprint or providing a specific position in every step.


In contrast to a construction based on predefined instructions (plan execution method) using blueprints, the proposed approach would be effective for dynamic scenarios where the conditions of the sites are subject to continuous environmental changes and require gradual growth over time. For example, the system can sense the adjacent on-going constructions gradually obstructing and changing the lighting condition of the site and flexibly create instructions for the next growth to optimize the amount of solar radiation to each unit. Adding a robot arm with cognitive capabilities – a sensing robot arm with devices such as a real-time camera feed or an active construction module such as a sensible brick can be viable options for future explorations, as they can concurrently generate instructions, based on the current state of the system, to spontaneously adapt to change its goal for globally optimal performance. The paper listed below further explored how such implementations can be applied to architectural design and speculates as to the possibilities of open frameworks for design using computational methods.


Related Publications:


A Generative Approach to Robotic Fabrication

Narahara, T., Proceedings of the 31st eCAADe Conference, TU Delft: Delft University of Technology, Delft, Holland, vol. 1, 2013, pp. 673-68.


bottom of page