LLM + RAG Robotic_Arm Generator (Grasshopper C# STEP_00–STEP_09)
Wickerson Studios Product Summary
Robotic Arm — LLM + RAG Generator (Grasshopper C# STEP Pipeline)
Build high-LOD robotic arm CAD assemblies in Rhino/Grasshopper using a clean, modular STEP_00–STEP_09 node series. This system is engineered for real Grasshopper workflows: BasePlane-safe, performance-aware, and designed so your ArmPrompt can cause major, visible changes in geometry—without breaking the pipeline.
What it does
- Generates a complete robotic arm assembly: skeleton → joints → links → tool/end effector → panels → fasteners → sensors
- Optional LLM + RAG prompting converts your ArmPrompt + reference image + metadata into structured CFG/SFG overrides
- Includes an animation driver node so joints move with smooth trig cycles (great for exploration, demos, and render sequences)
- Finishes with export packaging + a QA metrics report (counts, validity, bounding box, warnings)
Why it’s different
- Not a toy prompt demo: it’s a production-style pipeline built the way serious GH systems are built—stateful, stable, capped, modular.
- Prompt actually matters: the ArmPrompt is translated into downstream geometry drivers (joint count, link proportions, joint housings, tool style, panel logic, sensor density, etc.) so changing prompt language changes the robot in a big way.
- Learnable architecture: each STEP is a clean lesson in state, parsing, kinematics, solid-building, placement, and export.
Perfect for
- Computational designers, digital fabrication studios, educators
- Anyone who wants a reusable “prompt → parametric assembly” template in Grasshopper
- Creating rapid robotic concepts for visuals, speculative engineering, or teaching
Get it at
https://www.wickersonstudios.com
README .txt (download + full text)
Download: https://www.wickersonstudios.com
0) WHAT THIS IS
This is a 10-node Grasshopper C# “STEP” pipeline that generates a high-LOD,
BasePlane-safe robotic arm CAD assembly (curves + Breps) from a Master CFG/SFG
and an AI-driven ArmPrompt workflow.
You get:
• A reproducible parametric robot-arm generator (joints, links, panels,
fasteners, sensors, tools) that stays stable in Grasshopper
• An optional OpenAI-powered “Prompt → CFG/SFG/RAG” step that makes the arm
radically change form based on the ArmPrompt + reference images + metadata
• Animation support (STEP_02a) to drive joint motion with smooth trig cycles
• Export packaging (STEP_08) and QA metrics reporting (STEP_09)
This is built as a learning product AND a production-friendly template.
- REQUIREMENTS
- • Rhino 8 + Grasshopper (recommended)
- • Grasshopper C# Script component (the “C#” node)
- • Internet + OpenAI API key (optional; only needed for STEP_01b)
- If you do not provide a key, the system falls back to deterministic
- generation so the pipeline still runs.
2) INCLUDED NODES (STEP MAP)
STEP_00 — INIT / MASTER CFG+SFG / EngineState
STEP_01b — OpenAI Prompting (ArmPrompt + Ref Image URL + metadata → CFG/SFG/RAG)
STEP_02a — ANIM DRIVER (internal time + trig patterns → JOINT_DEGS list)
STEP_02 — BasePlane-safe Arm Skeleton (joint frames + axes + curves)
STEP_03 — Links from Skeleton (3D link cylinders/solids + link centerlines)
STEP_04 — Joint Solids (revolute housing, caps, brackets; BasePlane-safe)
STEP_05 — End Effector / Tooling (tool breps + optional cutters + envelope)
STEP_06 — Panels + Fasteners + Sensors (procedural placement, proper offsets)
STEP_07 — Assembly / Optional Boolean Pass (lightweight merging + stabilization)
STEP_08 — Export & Packaging (dir/prefix, JSON/CSV, optional mesh/IFC hooks)
STEP_09 — QA Metrics / Report (counts, bbox, validity, warnings)
NOTE:
Some deployments choose to merge STEP_07 + STEP_08 depending on your needs.
3) QUICK START (10 MINUTES)
A) Create a new Grasshopper file in Rhino.
B) Drop 10x C# Script components and name them:
S00_ARM_INIT
S01B_ARM_OPENAI
S02A_ARM_ANIM
S02_ARM_SKELETON
S03_ARM_LINKS
S04_ARM_JOINTS
S05_ARM_TOOLING
S06_ARM_HARDWARE_PANELS
S08_ARM_EXPORT_PACKAGING
S09_ARM_QA_METRICS
C) Paste the matching script code into each C# component.
D) Wire the outputs to the next step’s inputs EXACTLY by name.
(You can use “Relays” and “Scribbles” to keep the canvas clean.)
E) Set STEP_00 Run=true and provide:
• Seed (int)
• cfg (string)
• sfg (string)
F) (Optional) STEP_01b:
• Provide your OpenAI API key
• Set ArmPrompt (string)
• Provide ReferenceImageUrl (string) and/or metadata text (string)
• Choose OverrideMode: merge (recommended), append, or replace
G) Turn on STEP_02a Run=true to animate the arm over time.
H) Preview Breps/Curves at each step (joints/links/panels etc.).
4) THE MASTER CFG (HOW YOU CONTROL GEOMETRY)
The pipeline uses a Master CFG string (key=value; per line or semicolon).
Examples of CFG keys you will typically control:
• robot scale & proportions
• joint count
• link lengths, radii, taper
• joint housing sizes
• panel spacing / thickness / jitter
• fastener patterns
• sensor densities / offsets
• LOD caps (segment counts, max counts)
• performance caps (time budget / batching)
Tip:
If you want BIG, visible changes, change:
• jointCount
• baseReach / linkLen[] profile
• jointAxis modes
• endEffector type
• panel enable/spacing/rotation
• LOD/segmentation (within reason)
5) SFG (STRUCTURED FOCUS GUIDANCE)
SFG is a “channel” text block:
reference: ...
program: ...
style: ...
massing: ...
facade: ...
math: ...
notes: ...
For robot arms:
• “massing” describes kinematic silhouette
• “math” describes joint patterns (e.g. sine-cycled, easing, clamping)
• “notes” describes manufacturing/assembly logic
6) ARMPROMPT (MAKE THE ROBOT CHANGE)
ArmPrompt is the primary creative control for STEP_01b.
Use it to specify:
• Arm type (6-axis industrial / cobot / surgical / SCARA / delta)
• Material language (aluminum castings / carbon fiber / anodized plates)
• Joint styles (large harmonic drive housings / slim bearings / exposed bolts)
• Cable routing style (external chain / internal channels / braided looms)
• End effector (gripper / spindle / welder / suction / tool changer)
• Performance intent (reach, payload, stiffness, speed)
• Aesthetic (brutalist, aerospace, toy-like, biotech, etc.)
If you provide a reference image URL, STEP_01b can translate it into CFG.
7) ANIMATION (STEP_02a)
STEP_02a is an internal time driver:
Inputs:
• Run (bool)
• Reset (bool)
• Tick (bool) — can be a Toggle True or Timer-triggered boolean
• Dt (double) — time step, e.g. 0.1
• UseInternalTime (bool)
Outputs:
• TIME_OUT (double)
• JOINT_DEGS (List<double>) — wire to STEP_02 joint degrees input
This creates smooth undulating motion via trig/easing modes.
You can pick per-joint phase offsets, amplitude, bias, and smoothing.
8) EXPORT (STEP_08)
Suggested defaults:
ExportDir:
C:\Wickerson\Exports\RobotArm
ExportPrefix:
ARM_{seed}_{yyyyMMdd_HHmm}
STEP_08 writes JSON/CSV manifests and any selected exports.
If you do not want disk writes, keep ExportDir blank / Run=false.
9) TROUBLESHOOTING
• “Nothing outputs”:
- Confirm Run=true for the step
- Confirm you wired the correct upstream outputs into the correct input names
- • “Empty Sensors/Panels/Fasteners”:
- Confirm enable keys in CFG (and that STEP_06 is Run=true)
- • “Too heavy / slow”:
- Reduce LOD caps: maxSeg, maxParts, fastenerCount, panelCount
- Turn off cutters/booleans first
- • “Parts are rotated / offset”:
- Ensure BasePlane is consistent (WorldXY recommended)
- Confirm STEP_02 uses the same plane used by downstream steps
10) LICENSE / USAGE
This system is provided for educational and production use as a template.
You are responsible for validating engineering safety, structural integrity,
and manufacturing feasibility for real-world fabrication.
WICKERSON STUDIOS — THE WICKERSON WAY
Computational design assets, Grasshopper C# pipelines, and LLM-RAG workflows: