AGV Implementation in Thai Factories: Cost & AMR Comparison 2026

How much value is created by an operator walking a cart across your shop floor? In Thailand, rising minimum wages and a tightening labour market have pushed AGV implementation from “something to look at someday” to an active line item in the capital plan for many manufacturers. Yet the two comments we hear most often are still “we don’t know what to decide first” and “the quotations differ so wildly that we can’t compare them.” This article lays out the decision criteria for vehicle selection, the real cost structure and the payback formula, layout design, safety standards, host-system integration, and the practicalities specific to Thailand — in a form you can take straight into an internal review meeting.

Why AGV Implementation Suddenly Adds Up in Thailand and ASEAN

Ironically, the factories that looked at automated guided vehicles ten years ago and passed are the ones with the most to gain from looking again. The inputs to that decision have changed materially.

Rising labour costs have moved the break-even point between people and machines

Thailand’s minimum wage was raised to a range of THB 337–400 per day effective 1 January 2025, and on 1 July 2025 the whole of Bangkok moved to THB 400 per day (sources: Career Link, JETRO). Rates have since been held, however: as of July 2026 there is still no nationwide flat rate, and the applicable level remains THB 337–400 per day depending on the province (source: Bangkok Shuho). A figure in the region of THB 600 has been floated in political debate, but that is a discussion point, not a decision, so for investment purposes the realistic assumption is something conservative — current levels plus roughly 3–5% per year. General operator wages have been tracking the minimum wage at around 3–5% annual growth, and the manufacturing labour shortage adds further upward pressure.

The point most often missed is that the correct input for an investment case is not the hourly wage but the fully loaded annual cost per person. That means base pay plus statutory contributions, overtime, attendance/meal/transport allowances, dormitory and shuttle-bus costs, uniforms, medical checks, and the recruitment and training cost that recurs every year in proportion to turnover. In roles like material transport, where retention is typically poor, these hidden costs frequently add 30–50% on top of base pay. Ask your finance team for your own actual figures and put those into the formula. Borrowing an industry average — or worse, a Japanese head-office average — can shift your payback conclusion by years.

What market data says about the adoption phase

The AGV market is forecast at USD 6.51 billion in 2026, rising to USD 9.63 billion by 2031, a CAGR of 8.15% (source: market research report listed on GII). Looking at AGVs and AMRs together, another forecast puts the combined opportunity at USD 22 billion with an installed base of 3 million units by 2030, and annual shipments of roughly 1.38 million units in 2030 (source: market research report listed on GII).

What growth in unit volume actually means for a buyer is falling unit prices from production scale, and more options to choose between. One analysis frames 2026 as the phase in which adoption spreads to smaller sites and to complex inter-process transport (source: newscast.jp). AGVs used to pay back only in large warehouses and enormous automotive lines — simple, repetitive, high-volume movement. Today the applicable range extends to fleets of a few vehicles, high-mix production, and sites with convoluted routes.

Across industrial robotics as a whole, the IFR’s World Robotics 2025 reports 542,000 new industrial robot installations worldwide in 2024, roughly double the level of ten years earlier, with the regional split being 74% Asia, 16% Europe and 9% the Americas (source: IFR). The next edition, World Robotics 2026, is scheduled for release in autumn 2026. The fact that Asia is the main theatre of global automation investment is good news for factories in Thailand in a very concrete way: parts availability and the supply of engineers who have actually commissioned these systems.

Inter-process transport is usually the last manual task standing

In most plants, automation arrives first at the value-adding steps — machining, assembly, inspection — and the movement between them stays manual. Because transport looks like work “anyone can do”, it drops down the investment priority list. In practice it generates losses like these:

• A skilled operator leaves a machine to fetch parts, and the machine stops while they are away
• Late deliveries create line waiting time, which becomes chronic because the cause is never made visible
• Product damage from cart collisions, and pedestrian-versus-forklift risk in shared aisles
• “Let’s just move extra while we’re here” behaviour, which inflates work-in-process and eats floor space

Where transport should sit within your overall automation roadmap is covered in our article on the big picture of factory automation in Thailand. This article is the hands-on instalment focused specifically on transport.

AGV vs AMR: The Difference and How to Choose

If your team has not agreed on this distinction before quotations arrive, the internal discussion will go in circles.

AGV vs AMR at a glance

“AMRs are newer, therefore better” is a misreading. For high-volume repetitive transport on a route that will not change for three years, a magnetic-tape AGV is usually cheaper, more stable, and troubleshootable by your own maintenance team without a laptop. Conversely, choosing tape guidance for a site where the product mix turns over every quarter and aisles are shared with people and hand carts will fail operationally — you will drown in re-taping work and unplanned stops.

Comparing guidance methods

Hybrid approaches work well in the real world. Use SLAM in aisles shared with people, then add QR codes or a mechanical guide only in front of the machines where precise docking is needed. Handover to equipment sometimes demands accuracy of a few millimetres, and trying to achieve that through the navigation method alone sends the cost curve vertical.

Matching the vehicle format to the job

A seven-question decision flow

1. Are the weight, dimensions and shape of the load fixed or variable? Size for the 90th-percentile measured value, not the theoretical maximum
2. How many times per day does this move occur? Count peak and average separately
3. Does the route include steps, ramps, shutters, lifts or outdoor sections?
4. Are the aisles shared with people, forklifts and hand carts?
5. Is a layout change planned within three years?
6. Will loading and unloading be automated or done by a person? (This is where the cost really swings)
7. Is the vehicle called by a person pressing a button, or automatically by a host system?

Questions 1–3 essentially fix the vehicle format and payload, 4–5 decide AGV versus AMR, and 6–7 define the peripheral equipment and integration scope. Put these seven answers on a single sheet before you talk to vendors and the comparability of the quotations you receive improves dramatically.

Five Things to Decide First in an AGV Implementation

1. Define the load

“We move parts” is not a specification. Build a list by part number: weight, external dimensions, packaging format (cage cart / pallet / container), stack height, centre-of-gravity offset, and collapse risk. Centre-of-gravity offset and tolerance to vibration in transit are the items that turn into incidents. If you handle precision components or liquids, the permissible acceleration and deceleration limits will drive the vehicle choice.

2. Takt time and fleet size

Fleet size is calculated, not guessed.

Cycle time (min) = round-trip distance (m) ÷ effective speed (m/min) + loading time (min) + unloading time (min) + waiting time (min)

Vehicles required = (daily number of moves × cycle time (min)) ÷ (daily operating time (min) × availability factor)

Note: convert daily operating time into minutes before substituting (e.g. an 8-hour shift = 480 minutes). Leaving it in hours understates the fleet size by a factor of 60.

Three practical notes:

• Effective speed is 50–70% of catalogue top speed. Acceleration and braking, slowing for corners, temporary stops from human detection and yielding at intersections all happen, every day
• Start with an availability factor of 0.7–0.85. It absorbs charging, idle waiting and recovery from minor faults
• If the answer is 3.2 vehicles, consider starting with three and covering the peak with a person rather than buying four. You can always add vehicles; you cannot un-buy one

3. Layout and traffic flow

We cover this in detail below, but at this stage the question to answer is how much dedicated AGV aisle you can realistically secure. Very few factories can dedicate the entire route; shared sections almost always remain. The length of those shared sections and the number of intersections translate directly into safety design and cost.

4. Rules for coexisting with people

This is an operations problem, not a technology problem. Who responds when a vehicle stops? What rule — not a penalty — do you write for the moment someone puts a pallet on the AGV path? Who covers the night shift? Skip this and you create the classic outcome: everyone walks past a stopped AGV and nobody touches it.

5. Scope of host system integration

Will a person press a button, or will the production control system or WMS issue the instruction automatically? Aiming for full automation from day one is rarely wise; phase it. More on this below.

AGV Cost Benchmarks and Total Cost of Ownership

Vehicle price benchmarks (Japanese domestic market figures)

The figures below are published benchmarks for the Japanese domestic market. Do not apply them directly to a project in Thailand: import duty, freight, installation, local engineering labour and support costs including interpretation are all additional. Prices also vary substantially with exchange rates and with the sourcing route (Japanese, Chinese, European or locally assembled in Thailand). Approximate USD equivalents are shown at roughly JPY 150 = USD 1, for orientation only.

Search “AGV price” and you get vehicle prices. In an actual project the vehicle is roughly half the total, and under some site conditions considerably less than half.

Costs beyond the vehicle

RaaS and leasing

Robotics as a Service (RaaS) is reported to start from around USD 1,500–3,000 per robot per month (source: Modern Materials Handling). In the Japanese domestic market, a five-unit lease benchmark is around JPY 150,000–400,000 per month.

If you are considering RaaS in Thailand, get two questions answered at contract level: who arrives, within how many hours, when a vehicle fails, and whether spare units are physically held in-country. Actual support capability across ASEAN varies considerably between providers.

How to Think About Return on Investment

The payback formula

Simple payback (years) = total initial investment ÷ (annual savings − annual running cost)

• Total initial investment: vehicles + peripherals + construction + integration development + training (less any BOI import duty exemption you obtain)
• Annual savings: headcount reduced × fully loaded annual cost per person + secondary effects
• Annual running cost: maintenance contract + consumables + electricity + the labour equivalent of the management effort

Counting headcount reduction honestly

The most common error is reasoning like “transport takes three hours a day, so that’s 0.4 of a person.” Unless those three hours are converted into other value-adding work, headcount does not fall. The reductions you can credibly book are these:

• Eliminating a whole position (three people on a three-shift pattern, two on a two-shift pattern)
• Avoiding a planned hire (meeting a volume increase without adding people)
• Genuinely reducing overtime hours
• Genuinely reducing agency or subcontract labour

A useful sanity check is to convert everything into a cost per move. Divide your measured annual transport labour cost by the measured annual number of moves, then compare it with the annualised cost of the automated equivalent (investment depreciated over its expected life, plus running cost, divided by the same number of moves). If the two figures are within a few percent of each other, the project is too marginal to survive the first assumption that turns out to be wrong, and the honest conclusion is either to widen the scope so the same vehicles carry more moves, or to wait.

In Thai factories, the “one position on three shifts equals three people” arithmetic is the single biggest driver of payback. Conversely, on a process running only one shift, the same investment can take close to three times as long to pay back.

Do not overstate secondary benefits

Lower work-in-process, less product damage, reduced injury risk and shorter lead times are all genuinely valuable. But in a capital request, present the payback on the core labour effect alone first, and show secondary benefits separately. If secondary benefits are your primary justification and the numbers do not materialise, your next investment request will not be approved.

A sensitivity table

Holding the total investment constant, always tabulate how payback moves when your assumptions move. The values below are an illustrative example to show the method, not results from a real project. Substitute your own figures.

The key discipline is to size phase one so that it still survives the pessimistic case. Most failed projects trace back to a single decision: approving a plant-wide rollout in one go, using the optimistic numbers.

Variables specific to Thailand

• Currency: whether the contract is denominated in THB, JPY or USD
• Whether BOI incentives apply (see below)
• Electricity: with 24-hour operation, charging cost is not negligible
• Local service response time: multiply it by your cost of downtime per hour to evaluate it properly

Practical AGV Layout Design

Aisle width and turning radius

Vehicle width plus side clearance is not enough. Add the rear overhang swing during turns, the cut-in of towed carts, and any overhang of the load beyond the vehicle body. As a rule of thumb, a practical starting point is vehicle width plus roughly 600–800 mm, with additional margin on sections where people walk alongside. Cut this back and staff end up flattening themselves against a wall every time a vehicle passes, which builds resentment quickly.

While you are walking the route with a tape measure, record two numbers that rarely appear in a specification but strongly influence the outcome: the percentage of the route length that is shared with people and forklifts, and the number of points where the vehicle has to cross a main aisle. Those two figures predict how much safety equipment you will end up buying, how often the fleet will stop, and how far your effective speed will fall below the catalogue figure. If most of the route is shared and crossings are frequent, the honest options are to redesign the route, to accept a slower cycle time in the fleet-size calculation, or to reconsider whether transport automation is the right first target for this area at all.

Floor checklist

• Joints with a step of 5 mm or more, expansion joints, drainage gratings
• Slopes (gradient, plus load shift and wheel slip going up and down)
• Floor wear, oil, water and swarf contamination, peeling resin coatings
• Rattling pit covers
• Water films on outdoor connecting corridors in the rainy season, and the indoor/outdoor illumination difference (for vision-based systems)

Floors are casually treated as “something we can fix later”, but repair requires stopping production, which makes it the single biggest threat to your schedule. Look at the floor on day one of the site survey.

Charging strategy

Put the charging station where it sits on the natural traffic flow with minimal detour, not simply where there happens to be free space. A long detour to charge quietly costs you several percentage points of effective utilisation.

Congestion and traffic management

As soon as you have more than one vehicle, intersections and opposing-traffic sections become an issue. Decide these at design time:

• Priority rules at intersections (first-come, or a designated priority lane)
• One-way sections (a longer one-way loop is frequently faster than two-way passing)
• Deadlock detection and automatic recovery (both vehicles yielding and neither moving)
• The threshold for adopting fleet management software (the benefit generally becomes clear above three to five vehicles)

Interfacing with doors, shutters and lifts

If the route crosses floors or clean-area boundaries, the existing equipment needs a signal interface added. Whether the original equipment maker will modify their kit, and how long they need, can determine the critical path of the entire project — so ask them at the concept stage, before anything else. Routes crossing fire shutters or fire compartments also require a check against fire-safety regulations.

Safety and ISO 3691-4

The standards to reference

For driverless vehicles and AMRs, the international reference is ISO 3691-4, “Industrial trucks — Safety requirements and verification — Part 4: Driverless industrial trucks and their systems”. The current version is the second edition, published in 2023 (sources: ISO, JQA, IDEC, FA-products).

The important thing is the change of philosophy introduced in the 2020 edition, which was the first edition. The earlier assumption was that workers would pay attention and stay out of the vehicle’s way. The current thinking is that the machine must guarantee safety even when a human makes a mistake. Related standards include ISO 13849-1 (safety-related parts of control systems), IEC 61508 (functional safety) and EN 1175 (electrical requirements for industrial trucks).

At Japanese-affiliated factories in Thailand, it is common practice to adopt ISO 3691-4 as an internal standard even where it is not directly mandated by Thai law — driven by head-office safety policy, insurance requirements and audit readiness. Asking vendors “how would you demonstrate conformity with ISO 3691-4?” is one of the fastest ways to expose the difference in engineering depth between bidders.

When to redo the risk assessment

This is the most important point and the most frequently overlooked. A risk assessment is not a one-off exercise performed at commissioning when tasks are defined and the map is built. It must be repeated whenever the software is updated or the system configuration changes.

In other words, safety is not a task you complete at go-live; it is a change management process that has to be embedded in your organisation. Who approves a change, who performs the re-assessment, and where the record is filed. Writing that procedure at implementation time is what determines the state of the system three years later.

Safety measures actually implemented on the floor

• Front and side laser scanners with two-stage zones: a slow-down zone and a stop zone
• Bumper switches (contact detection) triggering emergency stop
• Direction-of-travel indicators, audible warnings, beacons, projected light lines on the floor
• Mirrors, warning lights and a pause before crossing at intersections
• Emergency stop buttons on the vehicle plus an area-level stop capability
• Lockout/tagout procedures for maintenance work where people enter the operating area
• Acceleration and deceleration limits configured to prevent load collapse

One caveat on audible warnings: set the volume against the actual noise level of your shop floor. Quieter than the ambient noise and it is pointless; too loud and someone will disable it “because it is annoying.” A safety measure that can be disabled is not functioning as a safety measure.

Integrating AGVs with WMS, MES and Production Management Systems

Decide who “calls” the vehicle

The value of an AGV lies less in the driving itself and more in “the right thing arriving at the right place at the right time.” The heart of the design is therefore the trigger for the transport instruction.

The realistic path is to stabilise operations at L1, then extend to L2 and L3 process by process as the benefit is proven. Aiming at L3 across every process from the outset typically burns six months or more in specification alignment alone, by which time the shop floor’s enthusiasm has evaporated.

Capturing operational data

AGVs generate travel logs, so with a little design thought they become the first source of quantitative data you have ever had on inter-process lead time, waiting time and transport peaks. That is a substantial prize for the production control function.

• What moved from which process to which process, at what time
• Waiting time between arrival and the start of the next process
• Response time from call to arrival (which correlates directly with shop-floor frustration)
• Where stop events occur and why (a rich source of layout improvement ideas)

Cross-referencing that data with production results and process progress requires a receiving system on the MES or production control side. We set out that thinking in implementing a production management system in a Thai factory. For the division of responsibility with a warehouse WMS and inbound/outbound automation, see logistics DX and warehouse automation in Southeast Asia. This article deliberately focuses on in-plant work: inter-process transport, line-side supply and empty-container return, whose requirements differ from warehouse put-away and picking.

Integration spec items to nail down

1. Communication method (REST API, MQTT, OPC UA, PLC signals or CSV exchange)
2. Master data ownership (who maintains the destination location codes)
3. Fault behaviour (what the AGV does when the host does not respond, and vice versa)
4. Handling of duplicate and cancelled instructions
5. Recovery procedure after a communication outage, and how to fall back to manual operation
6. Log retention period and how logs are accessed

Items 3 and 5 never appear in the capital request, yet they have more influence on post-go-live stability than anything else. Confirm explicitly that the design lets the shop floor keep running when the host system goes down.

Practicalities of Implementing in Thailand

Using BOI incentives (as of 2026; confirm eligibility with the BOI for your specific project)

Incentives from the Thailand Board of Investment (BOI) have a material effect on the AGV/AMR investment case. The table below summarises information published as of 2026. Eligibility, conditions and durations differ case by case, so always confirm with the BOI or a qualified adviser for your specific project (sources: Pertama Partners, Alvarez & Marsal).

Two practical points. First, the machinery import duty exemption requires the paperwork to be done before importation — clear customs first and you may lose the benefit entirely. Check this before you lock in the procurement schedule. Second, the technology-upgrade category for existing BOI-promoted facilities is precisely designed for automation investments like AGVs. Do not rule yourself out on the assumption that “this isn’t new investment, so it won’t qualify” — it is worth asking. The broader picture of the scheme is also covered in our factory automation article.

Thailand is assessed as a market where robotics and smart manufacturing investment is concentrating, driven by the EV transition, the maturing EEC and the Thailand 4.0 policy agenda (source: Asian Robotics Review). That also makes it comparatively favourable within ASEAN in terms of engineering talent and parts availability.

Import and installation checklist

• Confirm HS codes in advance, and the timing of the BOI duty-exemption application
• Power specification (Thailand is 230 V / 50 Hz; Japanese-built equipment needs voltage and frequency compatibility checked)
• Type approval for radio equipment (NBTC — required for devices using Wi-Fi, LoRa and similar)
• Available production-stop windows for installation (long holidays such as Chinese New Year, Songkran and year-end are commonly used)
• Attendance by local engineers, and operation and maintenance manuals in Thai

Choosing a system integrator or vendor

Choosing on price alone creates pain after go-live. Score bidders against these criteria.

Quotation transparency deserves special emphasis. A single lump-sum figure cannot be compared with anything. Hand bidders the cost breakdown table above and ask for pricing at the same level of granularity.

The Situation in Vietnam

Vietnam’s warehouse automation market is estimated at around USD 900 million, with growth driven by e-commerce expansion and supply chain consolidation (source: Ken Research). Concrete developments reported include Viettel Post deploying AGVs and automated sorting, and Vietnam Post Logistics partnering with Phenikaa-X to combine AMRs with AI-based tracking. Government programmes are also described as promoting Industry 4.0 technology adoption in logistics and warehousing (source: VLR).

Compared with Thailand, the differences look like this.

What is running ahead in Vietnam is AMRs in logistics and e-commerce; in-plant inter-process transport is still at an earlier stage. For a group with sites in both countries, validating a configuration in Thailand first and then rolling the proven design out to Vietnam is a defensible sequencing strategy.

Common Failure Patterns

These are the typical ways projects stall, or systems quietly stop being used after go-live. Use it as a pre-implementation checklist.

One more worth adding: evaluating the system before it has had a chance to work. There is always a tuning period immediately after go-live, and judging on those numbers will lead you to declare failure prematurely. Agree the evaluation timing in advance — three months after go-live, for example.

Implementation Steps and Realistic Timelines

The following is a general guide. Actual durations vary considerably with plant size, the condition of existing equipment and the scope of integration.

For a small pilot (one or two vehicles, existing layout unchanged), three to six months from concept to live operation is common. A full implementation involving host-system integration and construction work more typically runs nine to eighteen months. Because the BOI application and import lead times drive the schedule, do not plan around the manufacturing lead time alone.

One final note: skip phase 1 and you lose the ability to prove the benefit later. A stopwatch, a step counter and one week of actual move counts. That unglamorous exercise is the strongest material you will ever have for getting the capital request approved.

Frequently Asked Questions

What is the difference between AGV and AMR?

An AGV follows a predetermined path defined by magnetic tape, markers or similar infrastructure. An AMR uses localisation technology such as SLAM to reference a map and decides its own route to the destination. The practical difference that matters most is behaviour at an obstacle: an AGV generally stops and waits, while many AMRs re-route around it and keep going. Sites with fixed routes and infrequent layout changes tend to suit AGVs; high-mix low-volume sites with frequently changing layouts tend to suit AMRs.

How much does AGV implementation cost?

Published benchmarks for the Japanese domestic market put a general AGV at JPY 2.0–5.0 million per unit, a compact magnetic-tape model at around JPY 2.0–3.0 million, an AMR at around JPY 1.5–5.0 million, and an automated forklift (AGF) at around JPY 15 million. Those are Japanese market figures; in Thailand, import duty, freight, installation and local support are additional. A complete system including route installation, fleet control software and safety equipment can reach tens of millions of yen. Do not budget on vehicle price alone — request a quotation itemised to include all the peripheral costs.

How long does an AGV project take from start to go-live?

A small pilot typically takes three to six months. A full implementation involving host-system integration or floor works more commonly takes nine to eighteen months. In Thailand specifically, the BOI application process and the manufacture, shipping and customs clearance lead times tend to define the critical path. Since construction has to fit into windows when production can stop, planning backwards from long holidays such as Chinese New Year or Songkran is the realistic approach.

Can AGVs be implemented in an older, legacy factory?

In most cases yes, but the pre-checks are more extensive. The main items are floor steps, slopes and joints; aisle widths; the signal interfaces of existing shutters and automatic doors; electrical capacity; and the Wi-Fi environment. Floors matter most because repair requires stopping production, which affects both cost and schedule. Even in an older building, an approach that limits scope to one or two processes and starts with SLAM-based AMRs — which need minimal floor works — is often practical.

How many vehicles do you need before an AGV makes sense?

Judge by the unit of labour you can eliminate, not by vehicle count. Even a single vehicle can pay back if it lets you eliminate one position that is currently staffed across three shifts, because that is three people. Conversely, five vehicles produce limited savings if a person still has to stand by at each process to load and unload. Note, though, that when several vehicles share an area, fleet management software becomes necessary for traffic control, and its benefit generally becomes clear above three to five vehicles.

Should AGVs be integrated with WMS or a production management system?

Value does increase with integration, but you do not need full integration on day one. Stabilise operations first with call buttons on the shop floor, then move to automatic calls triggered by equipment signals, and only then extend to automatic instructions from the WMS or production control system. When you do integrate, make sure the specification defines what the AGV does when the host system goes down, and the procedure for switching to manual operation.

What safety standard should AGV implementations follow?

ISO 3691-4 is the primary international safety standard for driverless industrial trucks and AMRs; the current version is the second edition, published in 2023. The 2020 first edition shifted the underlying philosophy from relying on worker vigilance to requiring the machine to maintain safety even when a person makes a mistake; related standards include ISO 13849-1, IEC 61508 and EN 1175. The critical practice is repeating the risk assessment not only at implementation but every time the software is updated or the system configuration changes. Embed it as a change management procedure.

How should the benefit of an AGV be measured?

Measure a baseline before implementation, without exception. Record daily move counts, time per move, total labour hours spent on transport, waiting time from call to arrival, and line stoppage time caused by transport. After go-live, measure the same indicators using the same method. Because there is always a tuning period immediately after start-up, agreeing the evaluation timing internally in advance — three months after go-live, for example — allows a fair judgement.

Summary

Success in AGV implementation is determined less by vehicle performance than by how the decisions are made. The key points from this article:

• Rising labour costs and the labour shortage in Thailand have moved the break-even point for transport automation. But base the investment case on your own actual fully loaded cost per person
• AGV and AMR is not a question of which is better; it is a question of fit. The dividing lines are how often the layout changes and how much the aisles are shared with people
• Cost is not determined by vehicle price. Compare on total cost of ownership including peripherals, floor works, wireless infrastructure, integration development, training and maintenance. Japanese domestic market benchmarks cannot be asserted as Thai prices
• Payback swings dramatically on whether you can eliminate an entire position. On a three-shift pattern, the effect is tripled
• Treat ISO 3691-4 as an ongoing operating process with re-assessment at every change, not a one-off task at implementation
• Integrate with host systems in stages. Stabilise with human calls first, then extend
• In Thailand, the BOI machinery import duty exemption and the technology-upgrade category are material to the case (as of 2026; confirm eligibility with the BOI for your specific project)
• Most failures are operational rather than technical: items parked in aisles, inadequate training, and no defined performance indicators

If you want one concrete action for next week, take a single week of measurements on your current transport activity. How many trips, how many metres, how many minutes. Reviews that begin without those numbers usually stall halfway.

Even if you are still undecided about whether to automate at all, we are happy to talk through practical questions — how to capture your current transport data, or whether your particular processes point towards an AGV or an AMR. Working across both factory IT and FA in Thailand, our view is that transport automation is inseparable from getting production management and equipment data in order, and the earlier the whole picture is mapped out together, the fewer reversals the plan tends to need later. If you would like a conversation at the concept stage, our contact form is open.

References

• AGV market size and growth forecast: https://www.gii.co.jp/report/moi1939584-automated-guided-vehicle-agv-market-share-analysis.html
• AGV/AMR market opportunity and installed base forecast: https://www.gii.co.jp/report/logi1677406-agv-automated-guided-vehicles-amr-autonomous.html
• Analysis of the 2026 AGV adoption phase: https://newscast.jp/smart/news/2649660
• IFR World Robotics 2025 (new industrial robot installations): https://ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years
• Logistics robot cost benchmarks (Japanese domestic market): https://www.optimax.co.jp/ai-information/logistics-robot-cost/
• Commentary on AGV implementation cost (Japanese domestic market): https://logi.toyota-lf.com/column/20250324/
• Warehouse automation guide / RaaS pricing: https://www.modernmaterialshandling.com/automation/warehouse-automation-guide/
• Safety standard trends for driverless industrial trucks (JQA): https://www.jqa.jp/service_list/fs/robo_trend/02/index.html
• ISO 3691-4 (current edition: second edition, 2023) standard page: https://www.iso.org/standard/83545.html
• Explanation of ISO 3691-4 (IDEC): https://idec-fs.com/mir/column/column-4/
• AGV/AMR safety guidelines: https://fa-products.jp/column/agv-amr-safety-guideline/
• Thailand minimum wage guide 2025: https://kyujin.careerlink.asia/blog/thailand-minimum-wage-guide-2025/
• Thailand minimum wage held at current levels (2026): https://bangkokshuho.com/thainews-948/
• JETRO report on wage trends in Thailand: https://www.jetro.go.jp/biz/areareports/special/2024/0303/f5b4d6344434b2a9.html
• Thailand BOI incentives for manufacturing: https://www.pertamapartners.com/funding/thailand-boi-manufacturing
• Recent developments in Thai BOI incentives (2026-2027): https://www.alvarezandmarsal.com/thought-leadership/thailand-s-renewed-boi-incentives-a-strategic-window-for-growth-expansion-and-investment-2026-2027
• Robotics and smart manufacturing investment trends in Thailand: https://asianroboticsreview.com/home745-html
• Vietnam warehouse automation market: https://www.kenresearch.com/vietnam-warehouse-automation-market
• Vietnam’s factory automation turning point: https://vlr.vn/robotex-2026-and-vietnam-s-factory-automation-turning-point-29862.html