By Simon Bumford, Founder & CEO, Forge Robotics · Last updated: 2026-09-22
Are humanoid robots used in UK warehouses yet?
A handful are in pilots, but humanoids are the newest and least proven layer of warehouse robotics — most UK warehouse automation today is fixed or wheeled, not humanoid. The most-cited example is Agility Robotics’ Digit moving totes, publicly piloted with GXO Logistics; UK deployments specifically remain early-stage and trial-led.
That early status is the opportunity rather than a reason to wait. Warehousing is the single most active proving ground for humanoids globally, so a UK operator trialling now is following a well-mapped path and building operational know-how while competitors are still watching demo videos.
What can a humanoid robot do in a warehouse?
Humanoids suit repetitive, bounded handling tasks in human-shaped spaces — the jobs designed around people that fixed automation never economically covered. The realistic candidates UK logistics operators are exploring include:
- Moving totes and cases between stations, conveyors and racking
- Induction and decant — feeding goods into automated systems
- Palletising and depalletising support
- Container and trailer unloading (an emerging, high-interest use case)
- Replenishment and put-away of light items
- Put-wall, sortation and returns handling support
How do humanoids fit alongside the robots warehouses already use?
Humanoids complement, rather than replace, the autonomous mobile robots (AMRs), automated storage and retrieval systems, conveyors and robotic arms already common in UK warehouses. Fixed and wheeled systems remain far more efficient for structured, high-throughput movement.
Humanoids earn their place in the flexible, awkward, human-shaped gaps between those systems — the tasks left manual because they did not justify dedicated automation, or because the layout was built for people. The smart question is not "humanoid or AMR?" but "which task wants which machine?", and a manufacturer-neutral comparison answers it honestly.
Humanoid, AMR, fixed automation or people: which suits which warehouse task?
The table below is the short version of that comparison. Read each row for the task you have and let the machine follow from it; the full task-by-task version, with AGVs and the safety standard that applies to each, is in humanoid robots vs AMRs in UK warehouses.
| Humanoid robot | AMR | Fixed automation | People | |
|---|---|---|---|---|
| What it does best | Handling at human-height fixtures; tasks with no automation interface | Moving goods between mapped points; goods-to-person | Very high throughput on a fixed path: conveyors, sorters, AS/RS | Anything novel, variable or requiring judgement |
| Facility changes needed | Usually none | Mapping and charging points | Structural: racking, power, controls | None |
| Throughput character | Slower than a person per handling at first; no maker publishes a sustained rate | High for transport; limited by pick stations | Highest of all, only where it was built | The benchmark for every other column |
| Flexibility when the operation changes | High for task; low for speed | High for routes; moderate for load types | Very low | Very high |
| Maturity in UK warehouses (2026) | Pilots; early deployments reported by makers | Mainstream; reference sites and fleet software | Mature | The default |
| Commercial model | Pilot, then Robotics-as-a-Service or purchase, quoted per deployment | Purchase, lease or RaaS with fleet software | Capital project | Wages and agency rates |
| Safety framing | ISO 10218 / ISO/TS 15066; supervised, zoned | BS EN ISO 3691-4; shared-space rules | Machinery guarding; PUWER | Manual handling regulations |
| Good first task | Tote handling at a fixed station; decant into existing automation | Tote or cart transport between pick zones and packing | Not a first task; a designed system | Everything the others cannot yet do |
Qualitative comparison of the platform classes as offered to UK operators in 2026. No humanoid manufacturer publishes a sustained handling rate on its product pages as read on 22 September 2026, which is why a pilot measures it against the human rate on your own line.
Where do humanoids fit first in logistics?
The best first tasks are repetitive, physically tiring or hard to staff, and tolerant of a robot being slower than a person at first. Strong early candidates:
- Trailer and container unloading — physically punishing and hard to recruit for
- Tote handling between fixed points and conveyors
- Decant and induction into existing automation
- Repetitive replenishment of light, regular items
- Out-of-hours or peak-surge support alongside staff
Which humanoid platforms are being used for warehouse work, and what do their makers publish?
Fewer figures than the launch videos suggest. Every cell below is the maker’s own published position, read on 22 September 2026; “not published” means no official product page states it, and we have not filled the gap from press coverage or resellers. Boston Dynamics’ Stretch is included for comparison because it is the machine most often doing the job humanoids are being trialled for, and it is not a humanoid at all.
| Agility Robotics Digit | Figure 03 | Unitree G1 | Unitree H1 | Boston Dynamics Stretch (not a humanoid) | |
|---|---|---|---|---|---|
| Maker’s stated purpose | Tote handling, palletising and depalletising, machine tending, kitting and sequencing | “A general purpose humanoid robot for every day” | Humanoid platform for research, education and development | Full-size humanoid; “world record” walking speed claimed | Trailer and container unloading; case picking (wheeled base with an arm) |
| Form and size | Bipedal; dimensions not published on the product page | Bipedal; dimensions not published | Bipedal; 1,320 × 450 × 200 mm standing | Bipedal; about 180 cm | Wheeled base and arm; dimensions not published |
| Published weight | Not published | Not published | About 35 kg with battery | About 47 kg | Not published |
| Published payload | Not published | Not published | Arm load about 2 kg (EDU about 3 kg) | H1-2 arm load: rated about 7 kg, peak about 21 kg | Cases “up to 50 pounds” |
| Published runtime | Not published | Not published | About 2 hours (9,000 mAh battery) | Battery 15 Ah (0.864 kWh); runtime not published | “Long battery life” through “single or multiple shifts” |
| Sustained handling rate published | No | No | No | No | “Move hundreds of cases an hour” |
| Named deployments on the maker’s page | Amazon, GXO, Toyota Motor Manufacturing Canada, Mercado Libre, Schaeffler | None on the page as read; a BMW Spartanburg programme has been publicly reported | None named | None named | DHL, Maersk, Gap Inc., H&M, NFI, Arvato, Otto Group |
| Published price | Not published | Not published | US$13.5K (standard, unitree.com, 22 Sept 2026); EDU on request | Not published on the product page | Not published |
| UK availability as published | Not stated | Not stated | Via Unitree resellers | Via Unitree resellers | Not stated |
| Realistic UK route in 2026 | Pilot or Robotics-as-a-Service, quoted per deployment | Pilot; not openly available | Purchase; a research and development platform, not a warehouse worker | Purchase; as G1 | Purchase or per-deployment; in production at named customers |
Sources: agilityrobotics.com/products/digit (which notes that “Digit 5 is in development and the features, specifications, capabilities, and designs shown are subject to change”); figure.ai; unitree.com/g1 and unitree.com/h1; bostondynamics.com/products/stretch. All read 22 September 2026. The G1 and H1 are included because they are the humanoids UK buyers can most readily purchase, not because their makers position them for warehouse work.
Two things stand out. The enterprise humanoids aimed at warehouses publish customers but not specifications, and the humanoids that publish specifications are not aimed at warehouses. That is the state of the market, not a criticism of anyone, and it is why a UK operator’s first humanoid decision is a pilot with measures rather than a purchase from a table. For the one task where the makers do publish rates and customers, unloading, see trailer and container unloading robots in the UK.
What is still hard about warehouse humanoids?
Throughput, speed and battery life are the honest limits. A humanoid is generally slower and pricier per pick than purpose-built automation, charging or battery swaps interrupt continuous running, and reliability at warehouse uptime is still maturing. Grasping varied, deformable or poorly-presented items remains genuinely difficult.
So humanoids make sense where flexibility matters more than raw rate — variable tasks, awkward spaces, hard-to-staff shifts — not as a like-for-like replacement for a high-speed automated line. Judging them against the right benchmark is half of getting the decision right.
How do you keep a humanoid safe in a busy warehouse?
With the same discipline as any work equipment, sharpened for a busy, mixed-traffic environment: a documented risk assessment, defined working zones, and clear separation from people and materials-handling equipment such as forklifts. The framework is the Health and Safety at Work etc. Act 1974, PUWER 1998, and the ISO 10218 / ISO/TS 15066 robot-safety standards.
Early deployments run supervised and zone-limited, with a responsible person on site, rather than roaming a live operation autonomously. Planning that before the pilot starts is non-negotiable — a busy warehouse is an unforgiving place to improvise safety.
How many totes an hour can a humanoid robot move?
No manufacturer publishes a sustained figure, and any number you are given in a sales conversation should be treated as a demonstration rate rather than a shift rate. As read on 22 September 2026, neither Agility Robotics nor Figure states a handling rate for its humanoid on its product page, and Unitree positions the G1 and H1 as development platforms rather than warehouse workers. The only maker in the table above publishing a rate is Boston Dynamics for Stretch, which is a wheeled machine. The honest expectation for a first humanoid pilot is a robot slower than the person it works beside, improving over the pilot, and the right benchmark is the measured rate and error rate of that person at that station, not a conveyor and not a brochure.
What changed for warehouse humanoids in 2026?
Three things, all on the makers’ own pages as read on 22 September 2026. Agility Robotics now names Amazon, GXO, Toyota Motor Manufacturing Canada, Mercado Libre and Schaeffler on its Digit page, with Schaeffler described as planning to deploy humanoids across 100 plants by 2030, and states that Digit 5 is in development. Unitree lists the standard G1 at US$13.5K, below the figure quoted at its launch, which continues the downward pressure on the research-class price. And Boston Dynamics lists DHL, Maersk, Gap Inc., H&M, NFI, Arvato and Otto Group as Stretch customers, which is the clearest public evidence that the unloading task humanoids are chasing is already being done in production by a wheeled machine.
What has not changed is the absence of published specifications and rates for the enterprise humanoids, and the absence of a published UK deployment on any maker’s page. A UK warehouse trialling a humanoid in 2026 is still early, still learning, and still the right side of the curve to be on if the pilot is designed to produce a decision.
How to trial a humanoid in your warehouse
Start with a demonstration of suitable platforms against one defined task, then run a focused, supervised pilot with measures that matter in logistics — picks or cases per hour, accuracy, uptime, and hours reclaimed from hard-to-staff work. If it clears your threshold, scale on evidence; if not, you have spent little and learned exactly where the limits are. It is the same evidence-first process behind any sound UK robot deployment.
Frequently asked questions
Are humanoid robots used in UK warehouses?
A few are in pilots — Agility Robotics’ Digit moving totes (publicly piloted with GXO) is the most-cited example — but UK warehouse humanoid use is early-stage and trial-led. Most warehouse automation today is fixed or wheeled, not humanoid.
What can a humanoid robot do in a warehouse?
Repetitive, bounded handling in human-shaped spaces: moving totes and cases, induction and decant, palletising support, trailer and container unloading, light replenishment, and put-wall, sortation and returns support — supervised, alongside staff.
Are humanoids better than AMRs or fixed automation?
Not better — different. AMRs and fixed systems are more efficient for structured, high-throughput movement; humanoids suit the flexible, awkward, human-shaped tasks between them. The right answer is usually a mix, matched task by task.
What is the best first warehouse task for a humanoid?
Physically punishing, hard-to-staff jobs that tolerate a slower robot at first — trailer/container unloading, tote handling between fixed points, and decant into existing automation are strong early candidates.
What are the limits of warehouse humanoids?
Speed and throughput per pick, battery life and charging interruptions, reliability at warehouse uptime, and grasping varied or deformable items. They suit flexibility over raw rate, not replacing a high-speed automated line.
How do you keep a humanoid safe around forklifts and staff?
A risk assessment, defined working zones, separation from forklift traffic, and supervised, zone-limited operation under UK law (HSWA, PUWER) and the ISO 10218 / ISO/TS 15066 standards — planned before the pilot begins.
How do we trial a humanoid in our warehouse?
Demonstrate suitable platforms against one defined task, then run a supervised pilot measured on picks or cases per hour, accuracy, uptime and hours reclaimed. Scale only on the evidence.
Sources & references
- Agility Robotics — Digit humanoid for warehouse logistics
- UK Warehousing Association (UKWA)
- International Federation of Robotics — logistics and mobile robotics data
- Health and Safety Executive (HSE) — workplace transport and PUWER
- Agility Robotics — Digit product page (workflows and named deployments)
- Figure — Figure 03
- Unitree — G1 product page
- Unitree — H1 product page
- Boston Dynamics — Stretch product page
Related: humanoid robots for UK businesses · sectors we work in · how UK businesses deploy humanoid robots · which humanoid robots UK businesses can use · what humanoid robots cost in the UK · robot demonstrations for UK businesses · humanoid robots vs AMRs in UK warehouses · trailer and container unloading robots in the UK
Forge Robotics is an independent UK robotics market-entry partner. This article is general guidance and does not describe existing client relationships, live pilot programmes or any specific manufacturer.