Agent-Based Modeling

Agent-Based Modeling (ABM) is a powerful computational technique used to simulate the actions and interactions of autonomous agents to assess their effects on a system as a whole. Unlike traditional modeling methods that rely on top-down approaches, ABM focuses on the bottom-up design strategy.

The concept of ABM was pioneered by Craig Reynolds in the 1980s. Reynolds introduced the idea through his work on simulating the movement of flocks of birds, which he called "Boids." His model demonstrated how simple rules for individual agents (birds) could lead to the emergence of complex group behaviors. This foundational work has since influenced a wide range of disciplines, from biology to social sciences, and now plays a significant role in computational design.

ABM is a computational strategy that significantly differs from the linear logic often used in tools like Grasshopper. Instead of following a step-by-step process, ABM allows the designer to define a set of rules that govern the behavior of individual agents. These agents can represent many elements: a panel, a column, or a simple point guiding the creation of a path. The key is that they interact with one another and the environment, and from these interactions, complex forms and patterns emerge.

Agent-based modeling has found diverse applications across various fields, like urban planning, environmental comfort and escape route simulations.

Additionally, ABM has been utilized in architectural design to simulate the behavior of architectural elements. A notable example is the BUGA Wood Pavilion designed by the ICD & ITKE from the University of Stuttgart, where wooden panels were designed and arranged using agent-based principles, resulting in an innovative and structurally efficient form.

Diagram of BUGA Wood panel position and stresses. © ICD/ITKE University of Stuttgart

North view of BUGA Wood Pavilion. © ICD/ITKE University of Stuttgart

Another interesting application is the use of ABM in conjunction with robots to achieve Swarm robotics, also know as Distributed robotics. The agents are not only part of the design process, but they are also a physical entity that participates in the construction. An example of this is also from ICD & ITKE, the Research Pavilion 2014-15 where an industrial robotic arm is used to apply carbon fiber.

Agent-based design tool negotiating design parameters to determine fiber laying paths. © ICD/ITKE University of Stuttgart

Pneumatic formwork with robotic fiber reinforcement. © ICD/ITKE University of Stuttgart

The design process of the Life Lamp also took advantage of this strategy. You can read more about the Life Lamp publication here.

ICD released in 2022 their agent-based modeling framework, which evolved from almost a decade of research on this topic. It is called ABxM.Core, a plugin for Grasshopper.