Product definition
Prototype: Partial drawings and engineering knowledge.
Production: Released drawings, specifications, BOM and software baseline.
Robot Product Industrialization Guide
How to convert a functional robot prototype into a controlled, production-ready product
A prototype proves that one robot can work.
Production readiness requires a controlled product definition, repeatable manufacturing processes, qualified suppliers, stable calibration, validated tests and traceable configuration. This guide explains how to move a robotics product from an engineering prototype through design transfer, pilot production and manufacturing ramp.
Prototype to production is the controlled transition from a functional product prototype to a manufacturing system capable of repeatedly producing conforming units.
The transition includes design refinement, design for manufacturing and assembly, product-data release, supplier development, tooling, process definition, pilot production, calibration, testing, traceability and production ramp.
For robotics products, it must also control firmware, parameters, sensor calibration and robot-specific performance requirements.
See the parent robotics manufacturing lifecycle for the broader manufacturing-system context.
| Functional prototype | Production-ready product |
|---|---|
| Demonstrates technical feasibility | Demonstrates repeatable manufacturability |
| May use hand-selected components | Uses approved and controlled components |
| May depend on engineers | Uses documented production processes |
| May require manual adjustment | Uses defined calibration and limits |
| May contain temporary wiring or fixtures | Uses released production design |
| May use development firmware | Uses controlled production software |
| May not have serial traceability | Has unit-level configuration records |
| May be built slowly | Has realistic cycle-time evidence |
| May pass informal testing | Passes defined release tests |
| May ignore service and obsolescence | Includes lifecycle controls |
A prototype answers: can this concept work? Production readiness answers: can approved suppliers and controlled processes repeatedly build, calibrate, test and release it?
Prototype: Partial drawings and engineering knowledge.
Production: Released drawings, specifications, BOM and software baseline.
Prototype: Available or hand-selected parts.
Production: Approved manufacturers, approved suppliers and controlled alternates.
Prototype: Engineer-led assembly.
Production: Defined sequence, fixtures, parameters and work instructions.
Prototype: Manual tuning.
Production: Controlled calibration method, reference equipment and acceptance limits.
Prototype: Development build.
Production: Released firmware, parameters, keys, licences and recovery process.
Prototype: Demonstration testing and informal notes.
Production: Defined release tests and serial-level configuration history.
Design for manufacturing focuses on whether parts can be manufactured reliably and economically. Design for assembly focuses on whether those parts can be assembled efficiently, correctly and repeatedly. Design for manufacturing and assembly combines both perspectives at the product-system level.
| Method | Primary question |
|---|---|
| Design for manufacturing (DFM) | Can each part be manufactured under realistic process constraints? |
| Design for assembly (DFA) | Can the product be assembled efficiently and without avoidable errors? |
| DFMA | Does the complete design support efficient manufacture and assembly? |
| Design for test | Can required characteristics be measured and faults isolated? |
| Design for calibration | Can production variation be measured and compensated? |
| Design for service | Can critical parts be diagnosed and replaced? |
Design for manufacturing should not be treated as a final supplier check. Manufacturing constraints should influence the design before drawings, tolerances and tooling are frozen.
Related component interfaces are covered in the robot components, robot actuators and harmonic reducers guides.
Design transfer is the controlled transfer of the product definition and the knowledge required to manufacture, configure, calibrate, test and release the product. It is not simply the transmission of CAD files and a purchasing BOM.
The engineering BOM represents the product as designed. The manufacturing BOM represents the product as it will be sourced, structured, assembled and consumed during production.
| EBOM | MBOM |
|---|---|
| Organized around design architecture | Organized around manufacturing sequence |
| Owned primarily by engineering | Shared by manufacturing and operations |
| Represents intended product structure | Represents how the factory builds it |
| May omit consumables | May include adhesives, lubricants and packaging |
| May not reflect subassembly stations | Reflects production subassemblies |
| May not include manufacturing aids | May include process-specific items |
| Changes through design control | Must remain synchronized with approved design changes |
Robotics-specific MBOM fields may include robot model, variant, joint modules, controller configuration, sensor configuration, cable revision, firmware package, calibration package, country-specific components, optional end-effector interface and packaging configuration.
The MBOM must not become an uncontrolled alternative product definition. Every difference between the EBOM and MBOM should have a defined reason, ownership and change-control path.
A robot’s configuration includes more than its physical parts. It may also include firmware, controller parameters, sensor calibration, robot geometry, safety-related settings, licences, cryptographic keys and test-station compatibility.
provides configuration-management guidance across the product lifecycle from concept to disposal.
EVT, DVT and PVT are widely used product-development labels, but companies do not always use them identically. Each stage should therefore be defined through its evidence and purpose, not only through the acronym.
| Stage | Primary objective | Typical build state |
|---|---|---|
| Concept prototype | Prove the operating principle | Prototype components and methods |
| Engineering build / EVT | Validate architecture and major technical risks | Engineering-intent design |
| Design-validation build / DVT | Verify the released design against requirements | Near-production design |
| Production-validation build / PVT | Validate production processes and systems | Production-intent configuration |
| Pilot production | Demonstrate controlled repeatable manufacturing | Production-intent system |
| Ramp | Increase volume while maintaining control | Released production system |
A build should not be called PVT merely because it occurs late in the project. Its parts, tooling, operators, processes, software, calibration and tests must be sufficiently representative of production.
Pilot production is a limited manufacturing run used to validate the production system before larger-scale release. It should use production-intent components, tooling, processes, operators, documentation, software, calibration and test equipment closely enough to expose manufacturing risk.
Tooling should be developed when the product architecture and relevant interfaces are stable enough to justify investment, but early enough that production-intent tooling can be validated before ramp. Freezing tooling too early locks immature design decisions into the product. Waiting too long prevents pilot production from representing the real manufacturing process.
Tooling categories include machining fixtures, inspection fixtures, assembly fixtures, alignment fixtures, pressing tools, torque tools, cable fixtures, programming stations, calibration fixtures, end-of-line test fixtures, transport fixtures and service fixtures.
| Soft or interim tooling | Production tooling |
|---|---|
| Lower initial investment | Higher initial investment |
| Faster design iteration | Designed for repeatability and life |
| Useful for early builds | Suitable for representative cycle time |
| May require more manual adjustment | Should constrain critical variation |
| Limited capacity | Designed around expected production demand |
A work instruction should control an operation. It should not rely on undocumented engineer judgement to define whether the result is acceptable.
A robot assembled with the correct physical parts but the wrong firmware or parameters is not the correct product configuration.
See robot controllers for how software and motion control interact at the product-architecture level.
Prototype calibration may depend on manual judgement and engineering tools. Production calibration requires a controlled method, defined references, approved software, acceptance limits and unit-level records.
Calibration types include joint-zero calibration, encoder-offset calibration, robot-geometry calibration, force-torque sensor zeroing, camera-to-robot calibration, mobile-base calibration, payload configuration, tool-frame calibration and battery or energy calibration.
addresses measurement-management systems. addresses competence for testing and calibration laboratories. Related sensing methods are covered in robot sensors and machine vision.
Verification asks whether the design or process met its defined specification. Validation asks whether the resulting product or system meets its intended use under the defined conditions.
| Activity | Primary question |
|---|---|
| Design verification | Does the design meet the specified requirements? |
| Product validation | Does the product meet its intended use? |
| Process validation | Can the manufacturing process repeatedly produce the intended result? |
| Test-system validation | Does the test system correctly evaluate the required characteristics? |
| Unit release test | Does this particular unit meet release criteria? |
Do not imply that one final functional test substitutes for all five activities.
For industrial manipulating robots, is the relevant ISO reference for defined performance criteria and test methods; it should not be applied automatically to every service, mobile or humanoid robot. Safety scope depends on , and, where machinery risk assessment applies, .
End-of-line testing is the final production-stage verification that an individual unit has the correct configuration and meets its defined release requirements. It should detect unit-level defects. It should not be used as a substitute for controlled suppliers, capable processes or in-process quality controls.
See building a robotics production quality system for inspection and corrective-action frameworks.
First-pass yield is the percentage of units that complete a defined process and meet its acceptance requirements without rework, retest, repair or diversion.
First-pass yield = Units passing the process without rework ÷ Total units entering the process × 100%
Report the process boundary, build quantity, product revision, time period, excluded units, rework definition, retest definition and defect categories.
Final yield can look high when repaired units are counted as successful. First-pass yield exposes how frequently the process produces an acceptable result without intervention.
Do not use first-pass yield alone. Also review defects per unit, scrap, rework hours, retest rate, calibration failures, supplier defects, cycle time, escapes and field failures.
Production readiness is evaluated through evidence that the product, suppliers, manufacturing processes, tooling, people, measurement systems, tests, quality controls and capacity are sufficiently mature for the next production stage.
Manufacturing Readiness Levels provide a structured approach for assessing manufacturing maturity and manufacturing risk. This page may reference the framework, but does not claim that Yana’s checklist constitutes a formal MRL assessment.
Do not create an artificial overall percentage unless the scoring model, evidence weights and decision rules are publicly defined.
A contract manufacturer can support sourcing, production engineering, assembly, programming, calibration and testing. The appropriate point of involvement depends on the contract manufacturer’s engineering capability and on which product and manufacturing responsibilities the buyer intends to retain.
| Involvement stage | Potential benefit | Main risk |
|---|---|---|
| Early design | Manufacturing input before design freeze | Premature supplier dependency |
| Design-for-manufacturing stage | Process and tooling input | Advice optimized for one supplier |
| Design transfer | Manufacturing-system development | Missing product knowledge |
| Pilot production | Representative process validation | Late discovery of design problems |
| Serial production | Capacity and execution | Buyer loses process visibility |
Map responsibility for product design, BOM ownership, supplier approval, tooling ownership, manufacturing process, software provisioning, calibration method, test-system design, production records, engineering changes, quality responsibility, regulatory documentation and service and repair.
Contract manufacturing is an operating model. It does not remove the need to define product ownership, manufacturing responsibility and acceptance evidence. See qualifying a robotics supplier and assessing factory manufacturing capability.
| Pilot production | Production ramp |
|---|---|
| Validates the manufacturing system | Increases output through the validated system |
| Limited quantity | Increasing quantity |
| Exposes process weaknesses | Tests capacity and control under higher demand |
| Allows intensive engineering attention | Requires sustainable operating support |
| Establishes baseline yield | Stabilizes and improves yield |
| Validates traceability | Maintains traceability at volume |
Ramp controls include demand plan, supplier capacity, material availability, tooling capacity, assembly capacity, calibration capacity, test capacity, operator training, yield, rework, maintenance, quality staffing and escalation.
Production should not scale faster than the organization can contain defects, implement changes and preserve configuration control.
For cost modelling during scale-up, see robot manufacturing cost analysis.
| Failure | Likely consequence |
|---|---|
| Prototype BOM sent directly to factory | Temporary parts become production parts |
| Design for manufacturing performed after tooling order | Expensive redesign |
| Tolerances copied without process evidence | High cost or poor yield |
| Engineering knowledge remains undocumented | Factory depends on individual engineers |
| EBOM and MBOM diverge | Wrong parts or incomplete assemblies |
| Firmware not revision-controlled | Units behave differently |
| Pilot uses prototype tooling | Production problems remain hidden |
| Pilot is built by engineers | Operator and instruction issues remain hidden |
| Calibration relies on manual judgement | Unit-to-unit variation |
| End-of-line test developed late | Release criteria remain incomplete |
| Reworked units counted as first-pass yield | Process maturity is overstated |
| Supplier substitutions are informal | Performance and compliance risk |
| Production ramp begins with open critical failures | Defects scale with volume |
| Contract manufacturer owns undocumented knowledge | Switching risk |
| Field feedback is not linked to change control | Failures repeat |
China offers dense manufacturing ecosystems across machining, castings, moulding, motors, reducers, encoders, electronics, batteries, cables, sensors, robot joints and final assembly. This can shorten supplier discovery and iteration. It can also hide the division of responsibility between design owners, component factories, integrators, calibration providers and final assembly sites.
The transition therefore requires a product and responsibility map below the level of the supplier’s commercial name. Classify suppliers as complete robot OEMs, robot ODMs, contract manufacturers, electronics manufacturing services providers, mechanical component suppliers, joint or actuator-module suppliers, final assembly partners, calibration and test-system providers, system integrators, private-label suppliers or trading companies.
Ask which legal entity signs the agreement, which site performs final assembly, which processes are internal or outsourced, who owns the product design, firmware and source code, calibration algorithms, test software and fixtures, who approves component substitutions, who retains production records, who controls supplier changes, which components are imported, whether fixtures and test equipment can be transferred, whether another factory can reproduce the manufacturing process, and who supports overseas repairs and recalibration.
Map risk across design IP, process knowledge, tooling, firmware, calibration, test software, critical suppliers, documentation, production data, change control and service parts.
A factory capable of assembling a robot may not own the robot design, calibration method, control software or upstream component supply chain. The transition plan must identify both physical manufacturing capability and ownership of the knowledge required to reproduce the product.
It is the controlled transition from a functional prototype to a production system capable of repeatedly manufacturing, configuring, calibrating, testing and releasing conforming units. See the definition section.
Confirm the requirements, complete design-for-manufacturing and design-for-assembly reviews, release the product definition, develop suppliers and tooling, perform validation builds, execute pilot production and approve production readiness before ramping volume. See the twelve-stage process.
Not necessarily. A working prototype may rely on temporary components, engineering judgement, manual tuning and processes that are unsuitable for repeatable production. See prototype vs production-ready.
Design for manufacturing is the practice of designing a product with the capabilities, limitations, variation and economics of its intended manufacturing processes in mind. See DFM, DFA and DFMA.
Design for assembly focuses on making a product easier and more reliable to assemble by simplifying part relationships, access, orientation, fastening and error prevention. See DFM, DFA and DFMA.
DFMA combines design for manufacturing and design for assembly so individual parts and the complete assembly are optimized together. See the comparison table.
It should begin while product architecture, materials, processes and interfaces can still be changed without major tooling or redesign cost. See robotics DFM review.
Design transfer is the controlled movement of product requirements, drawings, BOMs, software, calibration, manufacturing methods and acceptance criteria from engineering into production. It is more than sending CAD files. See design transfer.
The EBOM represents the product as designed. The MBOM represents how the product is sourced, structured, assembled and consumed in production. See EBOM vs MBOM.
They are commonly used labels for engineering, design and production validation stages. Their exact definitions vary, so each stage should be defined by its inputs, production intent and required evidence. See EVT, DVT and PVT.
Pilot production is a limited manufacturing run used to validate production-intent components, tooling, processes, operators, documentation, calibration and tests before larger-scale production. Quantity alone does not define a pilot. See pilot production.
There is no universal number. The quantity should be sufficient to expose the relevant product, supplier, process, calibration, test and throughput risks for the intended production decision. See pilot build plan.
Production readiness is the demonstrated maturity of the product, manufacturing system, suppliers, tooling, people, tests, quality controls and capacity required for the next production stage. See the twelve readiness dimensions.
End-of-line testing verifies that an individual production unit has the correct configuration and meets its defined release requirements. It is not a substitute for process control. See end-of-line testing.
First-pass yield is the percentage of units that complete a process and meet requirements without rework, repair, retest or diversion. See first-pass yield.
Not necessarily. Final yield can include repaired or repeatedly tested units. First-pass yield, rework, defect distribution and process variation provide additional evidence. See first-pass yield.
A capable contract manufacturer can contribute during design for manufacturing, design transfer, tooling, pilot production or serial production. The timing depends on its engineering capability and the buyer’s intended ownership model. See contract manufacturing.
Define the product and responsibility model first. Then verify the actual production sites, internal and outsourced processes, critical suppliers, software ownership, calibration, test systems, traceability and transfer rights. See the China section and prototype-to-production service.
If you have a functional prototype and need controlled design transfer, DFM, pilot planning, manufacturing readiness or production-quality support, Yana can help structure the transition for robotics programs in China.