Tag: Agents

05
Feb

INTERVIEW- Daniel Bolojan

Founder of Nonstandardstudio [ 1 ], focused on development of innovative design methodologies and strategies, enacted through medium of computation, towards urbanism, architecture, design. Since 2014, Daniel Bolojan is a Junior Associate at Coop-himmelb(l)au in Vienna, where he has the leading role of developing custom computational tools, computational design strategies, virtual and augmented reality  applications, as well as robotic fabrication processes. There he had the opportunity to practice on numerous international projects and competitions   being leading the office push to develop.

He is currently a PhD Research Fellow at the University of Applied Arts Vienna, under the supervision of Patrik Schumacher, and a Lecturer at i.sd - Institute of Structure and Design -  Innsbruck University.

Images générées par un processus multiagent. ©Nonstandardstudio
Images générées par un processus multiagent. ©Nonstandardstudio
Images générées par un processus multiagent. ©Nonstandardstudio
Images générées par un processus multiagent. ©Nonstandardstudio

Could you please introduce yourself in few words?

Hello, I’m Daniel Bolojan, and I’m the founder of Non-Standard Studio. My main interests lie in creating complexly interrelated autopoietic systems, similar to natural systems, with subsystems capable of increased awareness, adaptability, towards all their components and towards their environment.

Can you share your definition of computational design with us?

Computational design involves that the design intent is set through a set of algorithmic instructions, of rules, a logic. Advantages of this approach are the designer's abitlity to access DNA source of structural and material behaviors, or of collective behaviors at multiple scales. These advantages translate into a more embedded design process where constraints and inputs of fabrication, social interaction, navigation, or environmental constraints are part and are driving forces of the design.

My work revolves around generative design and multi-agent algorithms. That is, understood in the studio’s work, a design process aiming at modeling the proto conditions of a system.

On appelle système multi-agent (SMA), un système composé d’un ensemble d’agents qui sont des entités réelles ou virtuelles, dont le comportement est autonome, évoluant dans un environnement, qu’ils sont capables de percevoir, sur lequel ils sont capable d’agir, et d’interagir entre eux.

Yves Demazeau, Populations and organizations in open multi-agent systems

 

Objet de longue date de recherches en intelligence artificielle distribuée, les systèmes multi-agents forment un type intéressant de modélisation de sociétés, et ont à ce titre des champs d'application larges, allant jusqu'aux sciences humaines.

How do you evaluate the quality of the result of this generative process?

First I have to point out that generative design follows a cycle through problem specification, design generation and design evaluation. Usually, while setting up the system, the designer can interactively control its behavior by specifying the design intent that needs to be embedded in the system. This design intent can represent constraints that must be met, desirable underlying logic it should follow, or characteristics the system should have. All these different types of design intents will have for sure different implications for how the generative system will work.

With regards to the evaluation of generative processes we have to keep in mind the two aspects of design evaluations.

On the one hand, the design intent's main features are embedded in the generative process which can directly result into possible evaluations of different quantifiable properties, characteristics of the design. Most of my work revolves around stigmergic models, composed of two parts: agents and the environment. The agents act according to pheromone trails stored into grid-like nodes that form the environment. The agents modify the environment  and get their behaviours modified according to the new modified environment. Behavioral characteristics are easily quantifiable and therefore could be subject to evaluations of the outcome by the system itself (e.g. find if an optimum path was found). In this case we can consider this a form of “system evaluation”.

On the other hand, however, there is a requirement for evaluations of emergent features of designs, that address as much technical reasons as aesthetic reasons. These evaluations are more difficult to compute from generative system’s design outcome, as they depend most of the time on subtle details of design aesthetics and context. For this type of evaluations there has to be a form of “designer evaluation”.

Could you explain the process you developed within “Ubiquitous Urbanism”?

Ubiquitous urbanism is a project that I did when I was studying in die Angewandte[ 2 ]/ Studio Zaha Hadid.

In the ecology of self-organized systems, there will always be different types of agent systems, with different behaviors and desires that will sense or modify their environment in a different way, based on their own characteristics.

At the urban scale, each main functional type (office, retail, recreation, housing, culture, transportation), had its own agent based system with its own functional type specifics/constraints. Series of interdependent differences, and correlation of resulting differentiated series, emerge according to the resulting stigmergy ecology, responsible for degrees of differentiation, different affiliation rules, and degrees of correlations.

The main idea of the project was to extract and use local rules of physical and visual connectivity as a bottom-up strategy to generate growth algorithms that will regulate spatial formations. These neighboring relations, based on physical and visual connectivity, don’t only allow a local neighboring negotiation, but also a global neighboring negotiation. The result is a coherent field not only on a local scale, but also on a global scale.

Ubiquitous urbanism (2013). ©Studio Zaha Hadid
Ubiquitous urbanism (2013). ©Studio Zaha Hadid

Our project explores principles of physical and visual connectivity as a method of evaluating and generating new spatial solutions for contemporary society. This idea derives from a research of individual urban systems, where on the example of working environment, we have addressed problems, needs and desires of corporative field."

Do you think your approach could be related to traditional urbanism?

If we mean by “traditional urbanism” the type of emergence that can be found in traditional urban formations as favelas, then my approach is very much so in line with traditional urban formations. As simple rules of emergence, such as neighboring rules, operate at an individual level and interact to give rise to the emergence of self-organization, a behavior emerges from these interactions and it’s something greater than the sum of its part.