Wiring Harness Drawing & Board Generation
Let AI Read Harness Drawings · Agent Exploration Prototype
Automotive harness drawings carry large volumes of non-standard, cross-platform, strongly-connected engineering information; going from drawing to production board diagram has long relied on manual reading, organizing, and drafting. We have built an "agent exploration pre-processing" prototype — it reads raw schematics directly, automatically identifies time candidates, geometric relations, label bounding boxes, and upstream/downstream connections, and outputs a structured Excel BOM. Understanding harness drawings is one of the key hard problems in industrial-drawing AI-ization.
Relationship With Other Systems
Each project is annotated with its position in the product system, making it easy to see at a glance how systems support each other.
Core Modules
Raw Schematic (Input)
Agent Exploration UI (Processing)
Excel BOM Extraction (Output)
Industry Pain Points
The harness industry has no shortage of CAD tools — what it lacks is a system that truly reads drawings, extracts BOMs, validates connections, and generates board diagrams.
Non-Standard Drawings
Customer, legacy, and cross-platform drawings vary in expression, hard to process automatically.
BOM Relies on Manual Work
Material info scattered across graphics, annotations, tables, and notes still needs manual organizing.
Complex Connections
Relations among wire numbers, holes, terminals, and branch paths exceed simple OCR.
Manual Board Drafting
Production board diagrams still largely depend on designers' manual understanding, layout, and correction.
If AI Could Read Harness Drawings
This is not efficiency optimization but a change in how work is done. Past: engineers read and draft. Future: AI understands drawings and auto-generates production results.
Why This Is Extremely Hard
This is not a drawing-recognition problem but an engineering-semantics understanding problem. Harness drawings simultaneously contain graphics, text, materials, connections, process rules, and production requirements.
Where the Industry Stands Globally
Many tools exist, but the key layer "non-standard drawing → AI understanding → production result" is missing.
| Type | Current Capability | Key Gap | Status |
|---|---|---|---|
| Int'l Harness SW | Design, BOM, mfg docs, assembly data | Strong in standardized design, weak at reverse-understanding non-standard drawings | Design-strong |
| Domestic CAD/CAPP | Drafting, process aid, BOM output, fixture board | Still highly dependent on engineer operation & judgment | Assist-strong |
| Engineering-Drawing AI | OCR, table recognition, dimension extraction, local understanding | Doesn't understand harness objects, connections, board craft | General-strong |
| Automation Vendors | Cutting, crimping, testing, assembly | Solve production actions, not upstream drawing understanding | Mfg-strong |
| What We Build | Drawing understanding, BOM, connectivity check, board generation | Fill the missing AI-understanding layer between drawing and production | Missing Layer |
Core Capabilities
Agent exploration pre-processing: auto-convert non-standard schematics into a structured Excel BOM — 9-step auto-exploration from data prep to entity comparison.
Raw Drawing Input
Accept customer non-standard schematics (CONTOUR/EXP/VT/RD/PD) without prior normalization.
Time-Candidate Filter
Filter 1s/3s/5s/10s time candidates from CONTOUR info via a filter.
Geometry Tooling
compute_stats calculates bbox medium-2/0/8/15/30 and more geometric params for later association.
Up/Downstream Validation
upstream_validation + xlh_bbox verify geometric and directory-structure consistency.
Multi-Variant Adaptive
default / adaptive / adaptive_light / adaptive_loose auto-run to fit different drawings.
Excel BOM Output
Output wire no., spec, wire type, color, left/right functional parts / terminals / seals — full fields.
Agent Exploration: From Schematic to Structured Data
The agent executes step by step — data prep → pre-refine → time candidates → geometry → association validation → directory gen → multi-variant → algorithm selection → application → verification — turning non-standard schematics into an Excel BOM.
Content Under Active Update
Product content has been published based on internal materials. The following areas are planned for further development:
Other Products & Solutions
Explore Xianma AI solutions in other domains