Trace Terra.

Can topographic, climatic, and biological data directly drive fabrication parameters — producing structures that are functionally responsive to their site and capable of hosting living biological soil crust?

Concretely: if we already know where the wind hits hardest, where the water collects, and where solar exposure peaks on a real piece of terrain, can those values be pushed straight into a robotic toolpath — so that every square metre of printed surface is tuned for the microbial community it needs to grow?

Biological soil crust (BSC) is a community of cyanobacteria, lichens, mosses, and fungi that grows across the top millimetres of dryland soil. A mature crust reduces wind erosion on horizontal surfaces by roughly 85%, binds sand into stable aggregates, and fixes both carbon and atmospheric nitrogen. It's effectively the skin of the desert — fragile, self-healing, and almost invisible.

Trace Terra's inoculation mix is tuned around three organisms: Nostoc commune for nitrogen-fixing cyanobacterial cover, Rhizobium spp. to support pioneer vegetation, and Actinomycetes to stabilise decomposed organic matter. Ratio 3 : 1 : 1, delivered in a growth medium that the printed surface can actually hold onto.

The pipeline starts with raw terrain: elevation meshes, slope and aspect rasters, wind-exposure and solar-irradiance maps derived from GIS and local climate datasets. Those inputs are imported into Grasshopper, where a parametric system decides four things for every location on the site — orientation, aperture, surface texture, and extrusion toolpath density.

Site-scale environmental behaviour — wind field, moisture flow, and rainfall erosion — was simulated end-to-end in Houdini, run across the full terrain mesh rather than on isolated geometries. Those simulation outputs fed back into the Grasshopper definition to refine placement and form. The resulting toolpaths can be produced at any resolution the robot can print: from single bricks for material testing up to full wall sections.

(Autodesk CFD enters later, at brick scale, for the thermal analysis of the printed diamond-infill geometry — covered in chapter 07.)

A single clay mix can't do everything the project needs, so we developed a three-layer wall. The outer layer is a porous clay-and-fibre blend tuned to host the BSC inoculation — rough enough for biofilm attachment, permeable enough to hold water long enough for microbes to use it. The middle layer uses an olive-pomace binder to create a hydrophobic retention boundary: water stays on the living face instead of draining through to the core. The inner layer is the structural workhorse — clay reinforced with hemp and palm fibres, set with alginate.

The three mixes were developed from a sequence of soil-substrate trays, testing how different fibre loads, binders (chitosan, alginate, methylcellulose, gelatine), and water ratios affected printability, dry strength, and biocrust retention. Each mix was then tuned for flow behaviour in the extruder.

The fabrication work ran across two robots at UCL Here East and the WASP 40100 delta printer. The primary production platform was a KUKA LBR iiwa running a custom pneumatic paste-extrusion end effector I designed and built: dual-pressure control so the outer and inner layers can flow at independent rates, an exchangeable cartridge, and a nozzle geometry matched to the three mixes.

I also set up and validated a second rig on a KUKA KR60 L45 — a 60 kg-payload, 2430 mm-reach industrial arm — to test whether the workflow would scale beyond the iiwa's build envelope. The rig hardware was built from scratch; the extruder malfunctioned mid-trials, which pushed the project back onto the iiwa but produced a working larger-format rig that subsequent cohorts can pick up.

The toolpath generator takes the Grasshopper geometry and turns it into robot-ready motion commands — infill density, spiral direction, layer-bonding dwells, and pressure ramps — all parameterised from the environmental model. Across iterations, setup inefficiencies dropped by roughly 40% as we standardised the nozzle-change and mix-prime routines.

Thermal performance was measured with a FLIR imager across solar-heating cycles. Under a 67 °C exposure, the diamond-infill brick held its interior surface roughly 3–4 °C cooler than a flat equivalent for the duration of the test — a small but consistent buffering effect that tracks the CFD predictions for airflow through the infill cavities.

Moisture-retention and biocrust-adhesion testing continued on soil trays in parallel, with weekly measurements of colony-forming units for the Nostoc inoculation. Early-stage growth confirmed that the outer-layer mix holds inoculant better than the structural mix on its own, validating the split-layer approach.

At full scale, the proposal is a cluster of terrain-responsive towers and low walls distributed across a selected Tabernas catchment — each element placed and shaped by the same Grasshopper model that drives its toolpath. Towers channel wind to shaded alcoves, walls terrace the slope to trap moisture, and every exterior face is a microbial host surface inoculated at the point of printing.

The masterplan doesn't pretend to re-green the desert. It defines a small, instrumented footprint where biocrust can be re-established measurably faster than on bare soil — a template that can be re-deployed across other degraded drylands by swapping only the input datasets.

The research has produced three academic outputs, led by the peer-reviewed IAAC BioDLF paper co-authored with Brenda Parker, Marcos Cruz, Anete Salmane, Jingyuan Meng, and Pradeep Devadass.