Introduction
In 2026, a small but growing number of humanoid robots are performing defined tasks in warehouses and manufacturing facilities. Some deployments have moved beyond short demonstrations into paid commercial agreements, while others remain pilots designed to determine whether humanoids can deliver reliable output at production scale.
The clearest examples are appearing in logistics and automotive manufacturing, where robots can be assigned repetitive physical workflows and their performance can be measured in operating hours, components handled, cycle times and throughput. The industry is therefore moving from “Can a humanoid robot do this?” to a more practical question: “Can it do the same job reliably enough to justify deployment?”
Where Are Humanoid Robots Actually Working?
Humanoid robot deployment is still concentrated in environments where the task is physically repetitive, the workflow is relatively structured, and the business can measure the robot's contribution.
1. Warehouses: Moving Totes Between Automation Systems
Warehousing is currently one of the strongest examples of commercial humanoid deployment.
Agility Robotics' Digit provides a useful example. At GXO's logistics facility, Digit works alongside existing autonomous mobile robots (AMRs) rather than replacing the warehouse's entire automation system.
The workflow is specific: AMRs bring product-filled totes to Digit's station, Digit unloads the totes and places them onto a conveyor that sends them toward downstream packing operations. When the conveyor becomes temporarily backed up, Digit can stack totes in another location and return to the conveyor when capacity becomes available.
This is considerably more concrete than simply saying that humanoids “move objects around warehouses.”
The deployment has also produced measurable operating data. Agility says Digit has moved more than 100,000 totes at GXO's Flowery Branch facility, including both transferring totes between AMRs and conveyors and stacking totes in another floor location.
GXO began regular operations with Digit in June 2024 under a multi-year agreement. The arrangement was structured as a Robot-as-a-Service (RaaS) deployment, meaning the business case is based on operational performance rather than simply purchasing a robot and installing it.
That makes logistics one of the clearest examples of humanoid robots moving from demonstrations into commercial operations.
2. Automotive Manufacturing: Handling Parts on Production Lines
Automotive plants are another important proving ground because they contain thousands of repetitive material-handling and assembly tasks.
BMW's Spartanburg plant in South Carolina provides one of the strongest documented examples.
In 2025, BMW deployed Figure 02 from Figure AI in an active production environment. Rather than performing an entire assembly operation, the robot was assigned a narrowly defined task in the body shop: retrieving sheet-metal components and positioning them accurately for welding.
Over the deployment period, Figure 02 operated for around 1,250 hours, worked 10-hour shifts on working days, moved more than 90,000 components and contributed to the production of more than 30,000 BMW X3 vehicles. BMW also reported approximately 1.2 million robot steps.
The significance is not simply that a humanoid “worked in a car factory.” It is that the deployment generated operational data at production scale.
BMW subsequently moved to a more complex application with Figure 03. Instead of repeating the original sheet-metal loading task, Figure 03 is being used for sequencing logistics. Figure says the robot's whole-body control allows its hands, arms, torso and feet to coordinate while it manipulates parts and pulls a heavy cart on caster wheels.
BMW is also testing humanoid robotics in Europe. At its Leipzig plant, the company is piloting AEON for applications involving high-voltage battery assembly and component manufacturing.
The progression is significant: first prove a repetitive task, then test whether a humanoid can handle a more complicated workflow.
3. Manufacturing Beyond Automotive
Humanoid deployment is also expanding into broader industrial manufacturing.
Schaeffler, a major German industrial supplier, has entered into an agreement with British humanoid robotics company Humanoid to explore deployment across its manufacturing network. The plan calls for an initial deployment at two German facilities from December 2026 through June 2027, including box-handling work at the Herzogenaurach site and broader testing at Schweinfurt. The longer-term agreement targets 1,000 to 2,000 humanoid robots across Schaeffler's global manufacturing sites by 2032.
This is important because it shows how companies are beginning to evaluate humanoids as part of a wider automation strategy rather than as isolated research projects.
The commercial objective is also clear. Manufacturers are interested in tasks that are repetitive, physically demanding or difficult to automate using fixed industrial machinery.
4. Factories Are Becoming the Main Testing Ground for Physical AI
Another notable deployment is Mercedes-Benz's testing of Apollo, a humanoid robot developed by Apptronik.
Mercedes-Benz has been testing Apollo under realistic production conditions at its manufacturing facilities and at its Digital Factory Campus in Berlin-Marienfelde. The objective is to understand how humanoid robotics can contribute to future production workflows.
This type of deployment should be described as a production test, rather than a full-scale autonomous workforce. That distinction matters because many humanoid robot announcements combine demonstrations, pilots and commercial deployments under the same label.
What Humanoid Robots Are Actually Doing
The most credible deployments show that humanoid robots are not currently taking over complete occupations. Instead, they are being assigned individual workflow steps.
These include:
- Loading and unloading totes from AMRs
- Placing totes onto conveyors
- Stacking containers when downstream capacity is unavailable
- Retrieving and positioning sheet-metal components
- Sequencing manufacturing parts for assembly
- Moving carts and material through production areas
- Handling boxes and other repetitive industrial materials
This task-level approach explains why humanoids are entering warehouses and factories before more complicated environments.
A company does not need a robot that understands every activity inside a warehouse. It needs a machine that can perform one defined operation repeatedly, safely and economically.
How Far Has Humanoid Robot Deployment Progressed?
The industry has clearly moved beyond the laboratory stage, but it has not reached mass autonomous deployment.
There are now several levels of maturity:
1. Demonstration
A robot performs a task under controlled conditions.
2. Pilot
A company places the robot in a real workplace to evaluate performance.
3. Operational deployment
The robot performs a defined task as part of a recurring workflow.
4. Commercial deployment
A customer pays for the robotic service, often through an RaaS model.
5. Scaled fleet deployment
Digit's GXO deployment and Figure 02's BMW production run are particularly useful because they provide evidence beyond a single demonstration.
At the same time, recent industry reporting shows that the gap between humanoid demonstrations and dependable commercial work remains significant. Many companies are still piloting robots, collecting training data and determining which tasks produce a viable return on investment.
The Biggest Barriers to Wider Deployment
The next phase of humanoid robotics will depend less on whether a robot can walk and more on whether it can deliver predictable economics.
1. Reliability: Factories need robots that can complete tasks consistently without frequent breakdowns or human intervention.
2. Battery Life and Uptime: Robots must operate for sufficient periods to support production workflows, while businesses need efficient charging and maintenance plans.
3. Safety: Humanoids working alongside people require reliable sensors, motion controls, emergency-stop systems, and facility-level safety measures.
4. AI Reliability: Vision-language-action systems help robots understand environments and adapt to unfamiliar situations, but complex real-world conditions can still require human intervention.
5. Cost Per Task: The business case depends on whether labor, energy, maintenance, software, integration, and supervision costs are competitive with human labor or traditional automation.
Finally, there is the question of cost per completed task. A humanoid robot does not create business value simply because it can perform a task. Its labor, energy, maintenance, software, integration and supervision costs must make sense compared with human labor or conventional automation.
Humanoid Robots Are Working, But the Real Test Is Scaling
Humanoid robot deployment has reached an important transition point. Robots such as Digit and Figure are no longer confined to laboratory demonstrations and have performed defined tasks inside live logistics and manufacturing operations. BMW's Figure 02 deployment generated measurable production data, while Digit's GXO deployment has crossed 100,000 totes. However, it would be premature to describe humanoids as replacements for human workers across entire facilities. The current model is more practical, with robots assigned to specific workflows where their performance can be measured through output, reliability, and efficiency.
The next major milestone will be moving from isolated tasks to multiple workflows, from pilots to larger fleets, and from supervised operation toward dependable autonomy. If manufacturers can achieve these milestones while controlling safety and operating costs, humanoid robots could become a meaningful part of industrial automation. For now, they are working primarily in carefully selected jobs where their physical flexibility can address specific operational challenges.
