Evolving permutations
Online, Zoom, Jun 4-5th and Jun 11-12, 2022
Introduction
The workshop will explore Genetic Algorithms as an exploratory and an optimisation process using Snowflake, a newly developed interactive plug-in, and apply the developed method to de- sign high-rise morphologies in a selected urban context.
Genetic algorithms are based on simulated biological evolution as a problem-solving machine. They are well known for solving multi-objective optimisation problems. Additionally, they have also been adopted by other creative domains such as art, music, fashion, industrial design and architecture. They are of particular interest to architects as they operate on populations, which are multiple individual designs, rather than just a single one. In design, they offer variational solutions related to the formulation of the design problem by each designer. As architectural design is a complex process that involves quantitative and qualitative objectives, architect’s intervention in the evolutionary loop is required to incorporate the designer’s subjective judgement.
Participants will be introduced to the working principles of Genetic Algorithms and run one non-interactive experiment in the first day. Participants will be invited to participate in three PhD experiments to test three levels of interactions with the algorithm in the 2, 3 and 4th day. Participation is voluntary. Full details will be provided after subscription.
METHODOLOGY
On the technical level, the workshop will introduce participants to the working principles of engaging with GAs to solve multi-objective optimisation problems first, then moving into the process of interacting with the algorithm to guide the evolution to reflect participant's intuition or preference. Using the newly developed interactive Snowflake plug-in, the learned design process will be applied in the generation of multi-story morphologies on a selected urban site. Participants will learn how to formulate the design problem as input data to the algorithm, and how to interact with the algorithm to guide the generation and selection process. The outcome of the design exercise will incorporate quantitative performance-based objectives, along with qualitative intuitive ones. The proposed process is an attempt to create a potential partnership between human intuition and the genetic algorithms that may enable a varied and richer space of viable design solutions.
/// Software & skills:
Basic modeling skill in Rhino, and intermediate to advanced level in Grasshopper are required. Participants should use their own laptop with pre-installed Rhino 7 software (software download links will be given after subscription).
// Tutors:
Zayad Motlib, Associate Professor XJTLU
Heba Eiz, Zaha Hadid Architect, Bartlett UCL
// Venue:
Online. Zoom links will be provided after registration.
// Calendar & Timetable:
The workshop will have the following timetable - Starting 1 pm GMT:
Jun 04- Introduction Lecture + Tutorials
Jun 05- Tutorials + Experiment 1a
Jun 11- Tutorials + Experiment 1b + 1c
Jun 12- Individual exploration of other interaction levels
// Subscription fees:
Free
// registration:
Email: unatworkshops@gmail.com
