Business Proposal

OpenMV
μ°¨μ„ΈλŒ€ 제쑰 ν˜μ‹ μ„ μœ„ν•œ
κ°œλ°©ν˜• λΉ„μ „-MES
μƒν˜Έμš΄μš© ν‘œμ€€ An Open Interoperability Standard
for Vision & MES
in Next-Gen Manufacturing

λŒ€ν•œλ―Όκ΅­ 제쑰 μ‚°μ—…μ˜ λ§‰νžŒ ν˜ˆκ΄€μ„ λš«μ–΄μ£ΌλŠ”
디지털 인프라 ν‘œμ€€ν™” 사업 Unclogging the arteries of Korean manufacturing through a
digital infrastructure standardization initiative.

1.1. 사업 λͺ… (κ°€μΉ­) 1.1. Project Name (working title)

OpenMV (Open Manufacturing Vision Interface):
이기쒅 제쑰 μ‹œμŠ€ν…œ κ°„ μ΄ˆμ—°κ²°μ„ μœ„ν•œ λΉ„μ „ 데이터 ν‘œμ€€ ν”„λ‘œν† μ½œ 및 고속 I/O 라이브러리 개발 사업 Developing a standard vision-data protocol and a high-speed I/O library for hyper-connectivity across heterogeneous manufacturing systems.

1.2. 좔진 λ°°κ²½ 1.2. Background

  • 제쑰 ν˜„μž₯의 '데이터 바벨탑' ν˜„μƒ: 핡심 ν’ˆμ§ˆ 검사 도ꡬ인 'λΉ„μ „ SW'와 생산 κ΄€λ¦¬μ˜ λ‘λ‡ŒμΈ 'MES' κ°„ μƒν˜Έμš΄μš©μ„± λΆ€μž¬. The "Tower of Babel" of factory data: there is no interoperability between Vision SW β€” the core quality inspection tool β€” and MES, the brain of production management.
  • ꡭ가적 λΉ„νš¨μœ¨ 심화: μž₯λΉ„ λ„μž… μ‹œ μ»€μŠ€ν„°λ§ˆμ΄μ§•(SI) λΉ„μš© κ³Όλ‹€ 및 λ¦¬μ†ŒμŠ€ λ‚­λΉ„ 초래. Compounding national inefficiency: every new piece of equipment racks up excessive SI customization costs and wastes resources.

1.3. 사업 λͺ©ν‘œ 1.3. Project Goals

  • "μ„ μ–Έν•˜λ©΄ μ—°κ²°λœλ‹€": λ³΅μž‘ν•œ μ½”λ”© 없이 라이브러리 호좜만으둜 μ¦‰μ‹œ μ—°λ™λ˜λŠ” ν‘œμ€€ ν”„λ‘œν† μ½œ 및 SDK 개발. "Declare it, and it connects": build a standard protocol and SDK that integrates instantly through a library call β€” no complex coding required.
  • K-Standard의 κΈ€λ‘œλ²Œν™”: κ΅­λ‚΄ 싀증 ν‘œμ€€ 확립을 ν†΅ν•œ κΈ€λ‘œλ²Œ 제쑰 데이터 μΈν„°νŽ˜μ΄μŠ€ μ£Όλ„κΆŒ 확보. Taking the K-Standard global: establish a domestically proven standard and lead the global manufacturing data interface.

2. ν˜„ν™© 및 문제점 (WHY) 2. Status & Pain Points (WHY)

νŒŒνŽΈν™”λœ 제쑰 μ†Œν”„νŠΈμ›¨μ–΄ μƒνƒœκ³„μ˜ ν•œκ³„ The limits of a fragmented manufacturing software ecosystem

λΉ„ν‘œμ€€ν™”λœ 연동 방식 Non-standardized integration

λΉ„μ „ SW κ°œλ°œμ‚¬λ³„ CSV, TXT, Socket, PLC Memory Map λ“± ν‘œμ€€ν™”λ˜μ§€ μ•Šμ€ λ°©μ‹μ˜ κ°œλ³„μ  μ‚¬μš© λ§Œμ—°. Each Vision SW vendor uses its own ad-hoc method β€” CSV, TXT, Socket, PLC Memory Map β€” with no shared standard.

데이터 톡합 λΆˆκ°€ No data integration

κΈ° λ„μž…λœ 타사 MES μ‹œμŠ€ν…œκ³Όμ˜ 톡합 운용 ν•œκ³„ 및 라인 증섀 μ‹œ 연동 λΉ„μš© 쀑볡 λ°œμƒ. Existing third-party MES systems are hard to integrate, and every line expansion duplicates integration costs.

λ°μ΄ν„°μ˜ 사μž₯ (Data Silo) Buried data (Data Silo)

λ‹¨μˆœ OK/NG νŒμ • μ™Έ AI μž¬ν•™μŠ΅μ— ν•„μˆ˜μ μΈ κ³ λΆ€κ°€κ°€μΉ˜ λ°μ΄ν„°μ˜ 미전솑 및 사μž₯(死藏). Beyond simple OK/NG verdicts, the high-value data essential for AI retraining never makes it out β€” and ends up buried.

κΈ°μ‘΄ ν‘œμ€€ν™” μ‹œλ„μ˜ ν•œκ³„μ™€ 차별점 How OpenMV differs from prior standardization attempts

κΈ°μ‘΄ κΈ€λ‘œλ²Œ ν‘œμ€€(OPC UA λ“±)은 'μ œμ–΄(Control)' μ€‘μ‹¬μ˜ 무거운 ν‘œμ€€μœΌλ‘œ, λŒ€μš©λŸ‰ 이미지와 AI 메타데이터λ₯Ό ν¬ν•¨ν•˜λŠ” ν˜„λŒ€μ  λΉ„μ „ μ‹œμŠ€ν…œ μ μš©μ— ν•œκ³„ 상쑴. Existing global standards (e.g., OPC UA) are heavyweight, control-centric specs β€” a poor fit for modern vision systems that handle large images and AI metadata.

ꡬ뢄 CategoryOPC UA / IPC-CFX (κΈ°μ‘΄existing)OpenMV (μ œμ•ˆ 기술proposed)
λͺ©μ  Purposeμ„€λΉ„ μ œμ–΄ 및 μƒνƒœ λͺ¨λ‹ˆν„°λ§ Equipment control and status monitoringλΉ„μ „ 검사 데이터(Rule-based + AI)의 포괄적 처리 End-to-end handling of vision inspection data (rule-based + AI)
방식 Approachλ³΅μž‘ν•œ λ…Έλ“œ ꡬ쑰 및 인증 절차 Complex node structures and certification flowsμ‹€μš©μ μΈ μŠ€ν‚€λ§ˆ(Schema) & 라이브러리(SDK) Pragmatic schema and SDK
λ‚œμ΄λ„ Adoptionλ„μž…μ— μˆ˜κ°œμ›” μ†Œμš” (높은 μ§„μž…μž₯λ²½) Takes months to roll out (high entry barrier)μ½”λ“œ λͺ‡ μ€„λ‘œ μ¦‰μ‹œ 적용 (Plug & Play) A few lines of code, instantly applied (Plug & Play)
접근법 PhilosophyTop-Down (μŠ€νŽ™ μ •μ˜ ν›„ κ΅¬ν˜„ κ°•μš”) Top-down: define a spec, then force implementationBottom-Up (SDK 배포λ₯Ό ν†΅ν•œ μ‹€μš©μ  ν™•μ‚°) Bottom-up: ship an SDK and let pragmatic adoption spread

3. 사업 λ‚΄μš© 및 기술적 μ ‘κ·Ό (WHAT) 3. Scope & Technical Approach (WHAT)

λ³΅μž‘μ„±μ˜ 좔상화 (Black-box SDK Strategy) Abstracting away complexity (a black-box SDK strategy)

핡심 μ „λž΅ Core strategy

톡신 ν”„λ‘œν† μ½œμ˜ λ³΅μž‘ν•œ λ‚΄λΆ€ ꡬ쑰(νŒ¨ν‚· μˆœμ„œ, λ°”μ΄λ„ˆλ¦¬ λ³€ν™˜, μ—λŸ¬ 핸듀링 λ“±)λ₯Ό κ°œλ°œμžκ°€ 인지할 ν•„μš” μ—†λŠ” μ™„λ²½ν•˜κ²Œ μΆ”μƒν™”λœ SDK(Black-box) ν˜•νƒœ 제곡.

- κ°œλ…: κ°œλ°œμžλŠ” "데이터 전솑" ν–‰μœ„μ—λ§Œ μ§‘μ€‘ν•˜λ©°, ꡬ체적인 전솑 λ‘œμ§μ€ ν‘œμ€€ λΌμ΄λΈŒλŸ¬λ¦¬κ°€ μ „λ‹΄ 처리.
- 지ν–₯점: USB μž₯치 μ—°κ²° μ‹œ λ“œλΌμ΄λ²„κ°€ μžλ™ ν†΅μ‹ ν•˜λ“―, SW λ ˆλ²¨μ—μ„œμ˜ Plug & Play κ΅¬ν˜„. Hide the gritty internals of the wire protocol β€” packet ordering, binary encoding, error handling β€” behind a fully abstracted, black-box SDK.

- Concept: developers focus solely on "sending data"; the standard library handles the transport logic.
- Goal: the way a USB driver auto-negotiates when you plug in a device, but at the software layer β€” true Plug & Play.

개발 λ²”μœ„ Scope of development

A. ν‘œμ€€ 데이터 μŠ€ν‚€λ§ˆStandard Data Schema (OpenMV Schema)

제쑰 데이터 ꡬ쑰λ₯Ό μ •μ˜ν•œ XSD 및 JSON Schema λͺ…μ„Έ λ™μ‹œ 제곡. 전톡적 수치/ν…μŠ€νŠΈ 기반 검사 결과뿐만 μ•„λ‹ˆλΌ, μ΅œμ‹  AI λͺ¨λΈμ˜ 벑터, 히트맡, μΆ”λ‘  신뒰도 λ“± λͺ¨λ“  검사 μœ ν˜• 수용 ꡬ쑰 섀계. We ship both XSD and JSON Schema specs that define the manufacturing data structure β€” accommodating not only traditional numeric/text inspection results, but also vectors, heatmaps, and inference confidence scores from modern AI models.

B. ν‘œμ€€ I/O SDKStandard I/O SDK (Reader/Writer Library)

Vision Side (Writer) 및 MES Side (Reader) 라이브러리 제곡. SW μ‹€μš©μ£Όμ˜(Pragmatism) 기반의 μ„ μ–Έν˜• API 섀계λ₯Ό ν†΅ν•œ 개발 생산성 κ·ΉλŒ€ν™”. Writer libraries for the vision side and Reader libraries for the MES side, with a declarative API grounded in software pragmatism β€” maximizing developer productivity.

기술 λ„μž… μ „ν›„ 비ꡐ (Code Level Simulation) Before / after, at the code level

[AS-IS: κΈ°μ‘΄ 개발 방식] [AS-IS: today's approach] λ³΅μž‘ν•˜κ³  였λ₯˜ λ°œμƒ κ°€λŠ₯μ„± λ†’μŒ Complex and error-prone
// Example: Manual Socket Implementation (Complex & Error-prone)
Socket client = new Socket(...);
client.Connect(ip, port);

// Manually constructing the packet string
string packet = "START" + "|" + result + "|" + imagePath + "|END";
byte[] data = Encoding.ASCII.GetBytes(packet);

client.Send(data);
// Need to handle reconnection, timeouts, parsing errors manually
[TO-BE: OpenMV 적용 방식] [TO-BE: with OpenMV] λ‹¨μˆœν•˜κ³  κ°•λ ₯함 (Simple & Robust) Simple & robust
// Example: Using OpenMV SDK (Simple & Robust)
// Declare the data structure
var inspectionData = new OpenMVData {
    Result = "NG",
    Image = currentImageBuffer,
    DefectCoordinates = new Point(10, 20)
};

// Send data using the standard library
// Connection and protocol handling are abstracted
OpenMV.Writer.Send(inspectionData);

4. 좔진 μ „λž΅ 및 μ‹€ν˜„ κ°€λŠ₯μ„± (HOW)4. Execution Strategy & Feasibility (HOW)

κ²€μ¦λœ μ—­λŸ‰κ³Ό μ‹€μš©μ£Όμ˜ μ² ν•™Proven capability, backed by a pragmatic philosophy

μˆ˜ν–‰ μ±…μž„μž μ—­λŸ‰Lead Engineer Track Record

λ³Έ 과제 μ±…μž„μžλŠ” ꡬ글(Google) μ•ˆλ“œλ‘œμ΄λ“œ 'WFF (Watch Face Format)'의 핡심 μ•„ν‚€ν…μ²˜ 단독 섀계(Sole Architect) 및 κ΅¬ν˜„ 이λ ₯ 보유.

λ‹¨μˆœ μ°Έμ—¬κ°€ μ•„λ‹Œ, XSD μ„ μ–Έ ꡬ쑰뢀터 검증 λ‘œμ§κΉŒμ§€μ˜ 전체 μ½”μ–΄ μ‹œμŠ€ν…œ λ…μž 개발 κ²½ν—˜μ„ λ°”νƒ•μœΌλ‘œ, 제쑰 μ‚°μ—… ν”„λ‘œν† μ½œμ˜ 기술적 완성도 보μž₯. The project lead was the sole architect behind the core architecture of Google Android's WFF (Watch Face Format) β€” not just a contributor, but the single designer and implementer.

That experience β€” building the entire core, from the XSD declarations down to the validation logic β€” is what guarantees the technical depth of this manufacturing protocol.

생산성 ν˜μ‹  μœ μΈμ±…Productivity Incentive

λΉ„μ „/MES κ°œλ°œμ‚¬μ˜ 톡신 λͺ¨λ“ˆ 직접 개발 λΆˆν•„μš”. 사업단 배포 κ²€μ¦λœ ν‘œμ€€ 라이브러리 ν™œμš©μ„ ν†΅ν•œ 즉각적 개발 착수 지원.

반볡적인 I/O μ½”λ“œ 개발 업무 제거 및 핡심 μ•Œκ³ λ¦¬μ¦˜ 개발 집쀑을 ν†΅ν•œ SW 생산성 비약적 μ¦λŒ€. Vision and MES vendors no longer need to build their own communication modules. The vetted standard library distributed by the project lets them start building immediately.

By eliminating the grind of repetitive I/O code and freeing teams to focus on the core algorithms, this delivers a step-change in software productivity.

싀증 및 ν™•μ‚° μ „λž΅Pilot & Adoption Strategy

μ¦‰μ‹œ 가동 κ°€λŠ₯ν•œ Testbed 확보: κΈ° 보유 제쑰 λ„€νŠΈμ›Œν¬(MES/λΉ„μ „ κΈ°μ—…, μ‹€μ œ 라인) ν™œμš©, 개발 μ¦‰μ‹œ μ•ŒνŒŒ/베타 ν…ŒμŠ€νŠΈ μˆ˜ν–‰.

Open Source μ „λž΅: 개발된 SDK의 GitHub λ“± μ˜€ν”ˆμ†ŒμŠ€ 곡개λ₯Ό ν†΅ν•œ μ§„μž… μž₯λ²½ 제거 및 자발적 ν™•μ‚° μœ λ„. A live testbed, ready on day one: we tap our existing network of MES vendors, vision companies, and real production lines to run alpha/beta tests as soon as the code is built.

Open-source distribution: we release the SDK on GitHub and other open channels to drop the entry barrier and let adoption spread on its own.

5. κΈ°λŒ€ 효과 (Impact)5. Expected Impact

κ΅­κ°€ 제쑰 μ‚°μ—… 및 제쑰 κΈ°μ—… 경쟁λ ₯ κ°•ν™”Stronger national manufacturing competitiveness

  • 제쑰 κΈ°μ—…(μˆ˜μš”μ²˜)의 톡합 관리 μ‹€ν˜„: ν‘œμ€€ λ„μž…μ„ ν†΅ν•œ νŠΉμ • λΉ„μ „/MES 업체 쒅속(Vendor Lock-in) νƒˆν”Ό 및 λΉ„μ „ 도ꡬ-MES μ„ νƒμ˜ μœ μ—°μ„± 확보 및 μ¦‰μ‹œ 톡합(Integration) ν™˜κ²½ κ΅¬ν˜„.
  • μŠ€λ§ˆνŠΈνŒ©ν† λ¦¬ 고도화 가속: 전사적 ν’ˆμ§ˆ 데이터 일원화 관리λ₯Ό ν†΅ν•œ κ΅­κ°€ 제쑰 μ‚°μ—… 전체 ν’ˆμ§ˆ μˆ˜μ€€μ˜ 상ν–₯ 평쀀화.
  • Unified management for manufacturers (the buy side): a shared standard breaks vendor lock-in to any single Vision/MES company, gives buyers flexibility in tool selection, and makes integration immediate.
  • Accelerated smart-factory maturity: consolidating quality data company-wide raises the overall quality bar across national manufacturing.

SW κΈ°μ—… 상생 및 κΈ€λ‘œλ²Œ μ§„μΆœSoftware vendor partnerships and global reach

  • λΉ„μ „ SW κΈ°μ—…: '연동' 이슈 ν•΄μ†Œ 및 'κ²€μΆœ μ•Œκ³ λ¦¬μ¦˜' 개발 집쀑 ν™˜κ²½ μ‘°μ„±, "International OpenMV Compliant" 인증을 ν†΅ν•œ ν•΄μ™Έ 수좜 μ‹ λ’°μ„± 확보.
  • MES κΈ°μ—…: λ‹€μ–‘ν•œ λΉ„μ „ μž₯λΉ„μ˜ μ†μ‰¬μš΄ ν”ŒλŸ¬κ·ΈμΈ(Plug-in) 지원을 ν†΅ν•œ ν”Œλž«νΌ ν™•μž₯μ„± μ¦λŒ€.
  • Vision SW companies: integration headaches go away, freeing teams to focus on detection algorithms. "International OpenMV Compliant" certification then becomes a credibility lever for export.
  • MES companies: easy plug-in support for a wide range of vision equipment expands platform reach.

6. κ²°λ‘ 6. Conclusion

λ³Έ 사업은 λ‹¨μˆœ 기술 개발 과제λ₯Ό λ„˜μ–΄, λŒ€ν•œλ―Όκ΅­ 제쑰 μ‚°μ—…μ˜ λ§‰νžŒ ν˜ˆκ΄€(Data Flow)을 λš«μ–΄μ£ΌλŠ” 디지털 인프라 ν‘œμ€€ν™” μ‚¬μ—…μž…λ‹ˆλ‹€. This is more than a technology development project β€” it's a digital infrastructure standardization initiative that unclogs the data flow of Korean manufacturing.

κ³Όκ±° ν•˜λ“œμ›¨μ–΄μ˜ ν‘œμ€€ν™”κ°€ μ‚°μ—… 혁λͺ…을 μ΄λŒμ—ˆλ‹€λ©΄, AI μ‹œλŒ€μ—λŠ” 데이터 ν”„λ‘œν† μ½œμ˜ ν‘œμ€€ν™”κ°€ μ‚°μ—… νŒ¨κΆŒμ„ κ²°μ •ν•©λ‹ˆλ‹€. κΈ€λ‘œλ²Œ ν‘œμ€€(WFF) μ•„ν‚€ν…νŠΈλ‘œμ„œμ˜ 성곡 κ²½ν—˜κ³Ό μ‹€μš©μ£Όμ˜ 철학을 λ°”νƒ•μœΌλ‘œ, ν•œκ΅­ν˜• 제쑰 ν‘œμ€€(OpenMV)의 κΈ€λ‘œλ²Œ 제쑰 μ‚°μ—… ν‘œμ€€(Global Standard) 도약 μ‹€ν˜„ν•˜κ² μŠ΅λ‹ˆλ‹€. Hardware standards drove past industrial revolutions; in the AI era, data-protocol standards decide who wins the industry. Drawing on the success of a global standard architect (WFF) and a pragmatic philosophy, we'll take the Korean manufacturing standard (OpenMV) and turn it into a global manufacturing standard.

Contact Us