Land Agents
LAB: TECH
DIVISION: DEVELOPMENT
By Chris Landau, former Manager of Design Technology at OLIN & Tech Lab Leader
Land Agents
Land Agents is a new approach to landscape planning that aims to synthesize analysis and design through new computational tools.
the problem with analysis
Site analysis is a means of understanding a site. But analysis is often a once-and-done proposition. Designers will do the required categories of analysis to 'understand' a site. Often, one or two bigger revelations come from these analyses. But the nuances of the data (if it is nuanced) are often lost in between the steps of analysis and design.
Typical layers of site analysis derived from Landscape Architect Ian McHarg's layer cake methodology.
Cognitive Limitations
The reason detailed aspects of analysis are lost can largely be attributed to the inability of the human mind to hold more than a few of them at once. To combat this, we are creating a reactive, agent-based platform that can help us to synthesize analytical layers at a finer level of detail. We also want to have ability to bring aesthetics, design assumptions, and less 'factual' layers of information into this synthesis. By including design assumptions and objectives in the process of design analysis, we can test objectives against analytic and assumptive layers.
Agent-Based
By creating a virtual analog of both analytical and aesthetic layers of a site, we can employ digital agents that explore this problem space within a simulation. Much like cells in a body find their way by following chemical gradients and through complex interactions with one another, land agents find desired conditions in a site based on the conditions and objectives provided. An agent may represent a specific site feature or plant but it may also represent programmatic zones or intensities. Unlike typical suitability analysis, agents form complex interactions. They also allow for multiple solutions to one set of inputs.
Home-Grown
Like many of our tools, land agents is coded in-house. It is written in Python and implemented in Rhino and Grasshopper, built on top of our SUPERSET tool set. We are exploring the many directions that these motile agents might carry us.
