CtrlK
| HMND 01 |
|---|
| General information |
| Manufacturer |
| Country of origin |
| Year introduced |
| Status |
| Variants |
| Height |
| Weight |
| Degrees of freedom |
| Payload capacity |
| Max speed |
| Battery life |
| AI framework |
| Compute |
| Pre-orders |
| Website |
HMND 01 is a modular humanoid robot developed by Humanoid (legally registered as SKL Robotics Ltd), a United Kingdom-based robotics and artificial intelligence company founded in 2024. The robot is designed for industrial automation tasks including goods handling, picking and packing, kitting, and logistics operations. It is available in two configurations: the Alpha Wheeled variant, which uses an omnidirectional wheeled mobile base, and the Alpha Bipedal variant, which features full bipedal locomotion for environments requiring stair climbing or navigation of uneven terrain.
The Alpha Wheeled version was unveiled in September 2025 and made its North American debut at CES 2026 in Las Vegas, while the Alpha Bipedal was announced in December 2025. Humanoid claims that the wheeled variant was developed in seven months and the bipedal variant in five months, making the HMND 01 one of the fastest-developed humanoid robots in the industry, where typical development cycles range from 18 to 24 months.[1][2] Both variants are powered by the company's proprietary KinetIQ AI framework, a four-layer software architecture built on NVIDIA computing technologies that enables autonomous task execution, fleet coordination, and integration with enterprise resource planning systems.
As of early 2026, Humanoid reports over 20,500 pre-orders for the HMND 01, has completed multiple proof-of-concept deployments at Siemens, Ford, and Martur Fompak facilities, and demonstrated multi-robot voice-controlled collaboration at NVIDIA GTC 2026.[3]
Humanoid was founded in May 2024 by Artem Sokolov, a serial entrepreneur and global investor who previously grew his family's jewelry manufacturing business to a $1 billion valuation.[4] Sokolov has described his motivation as deeply personal: watching his grandparents spend their entire working lives performing repetitive tasks in jewelry production inspired him to build robots that could "free people from routine and repetitive tasks" and enable them to pursue "more creative and meaningful work."[5]
The company is legally registered as SKL Robotics Ltd in the United Kingdom. Sokolov holds an Executive Certificate in AI and Digital Business Excellence from the IMD business school in Switzerland.[6]
Since its founding, Humanoid has grown rapidly. As of early 2026, the company employs over 200 engineers, researchers, and innovators, with team members drawn from leading global technology companies including Apple, Tesla, Google, Boston Dynamics, Sanctuary AI, and NVIDIA.[7] The company maintains offices in three locations:
Humanoid is backed by $50 million in founder-led capital. Unlike many robotics startups that rely heavily on venture capital, the company's initial funding comes primarily from Sokolov's personal investment, giving him significant control over the company's direction and development priorities.[8]
Humanoid operates on a Robots-as-a-Service (RaaS) model rather than selling robots outright. Under this model, enterprise customers lease HMND 01 units and pay recurring fees, which reduces the upfront capital expenditure required for deployment. The company claims this approach can deliver labour cost savings of up to 50% annually for customers.[9] Sokolov has indicated a target unit price range of $50,000 to $70,000 per robot, which translates to an effective operational cost of approximately $10 per hour, significantly below the average wage for equivalent manual labor in developed nations.[10]
Humanoid targets an addressable market of approximately 250 million workers across its initial focus sectors: retail, e-commerce, third-party logistics (3PL), manufacturing, and automotive. The company cites a Goldman Sachs projection that the humanoid robot market will reach $38 billion by 2035 and could grow to $1 trillion by 2050.[11] Humanoid has outlined a three-phase deployment strategy:
| Phase | Target year | Application domain | Addressable market |
|---|---|---|---|
| Phase 1 | 2027 | Physical tasks (manufacturing, warehousing, logistics) | 250 million workers |
| Phase 2 | 2029 | Service sector (elder care, hospitality) | 1.4 billion people over 60 |
| Phase 3 | 2031+ | Household applications | 3.5 billion households |
Humanoid unveiled the HMND 01 Alpha Wheeled in September 2025 as the UK's first humanoid robot designed for industrial use.[12] The wheeled variant was developed from initial concept to working prototype in seven months, a timeline that Humanoid attributes to its simulation-first development approach using NVIDIA Isaac Sim and Isaac Lab. By designing, testing, and training the robot extensively in simulation before building physical hardware, the engineering team was able to compress a development cycle that typically takes 18 to 24 months.[13]
The Alpha Wheeled stands 220 cm (7 ft 3 in) tall and weighs 300 kg (661 lb). It uses an omnidirectional wheeled mobile base for locomotion, reaching speeds of up to 2 m/s (7.2 km/h). The robot features 29 degrees of freedom (excluding end-effectors), a bimanual payload capacity of 15 kg, and an average runtime of 4 hours per charge.
The HMND 01 Alpha Wheeled made its North American debut at the Consumer Electronics Show (CES) 2026 in Las Vegas, held from January 6 to 9, 2026. At the Humanoid booth, the robot performed a live demonstration of autonomous bin picking, extracting metallic bearing rings from cluttered industrial bins in a near-production factory environment. The demonstration highlighted the robot's computer vision capabilities and manipulation dexterity under realistic operating conditions.[14]
On December 2, 2025, Humanoid announced the HMND 01 Alpha Bipedal, which went from first design to working prototype in five months.[15] The bipedal variant achieved stable walking just 48 hours after final assembly, a milestone that typically takes weeks or months for other humanoid platforms. This rapid sim-to-real transfer was made possible by extensive simulation training: the engineering team generated 52.5 million seconds of reinforcement learning locomotion data (equivalent to nearly 19 months of continuous training) in just two days using NVIDIA Isaac Sim and Isaac Lab.[16]
The bipedal Alpha stands 179 cm (5 ft 10 in) tall and weighs 90 kg (198 lb). It shares the same 29-degree-of-freedom upper body as the wheeled variant, enabling direct skill transfer between platforms. The robot's first real-world steps were taken after 3.2 million seconds of simulated training.
| Specification | Alpha Wheeled | Alpha Bipedal |
|---|---|---|
| Height | 220 cm (7 ft 3 in) | 179 cm (5 ft 10 in) |
| Weight | 300 kg (661 lb) | 90 kg (198 lb) |
| Degrees of freedom | 29 (excl. end-effectors) | 29 (excl. end-effectors) |
| Locomotion | Omnidirectional wheeled base | Bipedal legs |
| Max speed | 2 m/s (7.2 km/h) | 1.5 m/s (5.4 km/h) |
| Average runtime | 4 hours | 3 hours |
| Payload capacity (bimanual) | 15 kg (33 lb) | 15 kg (33 lb) |
| Vertical reach | Floor level to 2 m | Floor level to ~1.8 m |
| Shelf access depth | Up to 60 cm | Up to 60 cm |
| Processors | NVIDIA Jetson Thor | NVIDIA Jetson Orin AGX + Intel i9 |
| End-effectors | 12-DOF hand or 1-DOF gripper | 12-DOF hand or 1-DOF gripper |
| Battery | Swappable | Swappable |
The Alpha Bipedal demonstrates a diverse range of locomotion modes, all trained through reinforcement learning in simulation:[17]
The whole-body locomotion controller is trained entirely in simulation using online reinforcement learning, requiring approximately 15,000 hours of simulated experience to produce a capable model. The company reports that minimal domain randomization adjustments were needed for successful sim-to-real transfer.[18]
Both HMND 01 variants are equipped with a comprehensive sensor package designed for industrial environment perception:
| Sensor type | Details |
|---|---|
| RGB cameras | 360-degree head-mounted vision (6 cameras on Bipedal) |
| Depth sensors | 2x head-mounted depth sensors |
| Force/torque sensors | 6D F/T sensors on arms and end-effectors |
| Wrist cameras | RGB cameras mounted at each wrist |
| Haptic feedback | Haptic sensors for contact detection |
| Microphone array | 6-microphone array (Bipedal) |
| Additional sensors | IMU, gyroscope, temperature monitoring, ultrasonic sensors |
The HMND 01 uses a modular end-effector system, allowing operators to swap between two configurations depending on the task requirements:
This modularity extends to the robot's overall design philosophy. The hardware and software architecture allows customers to configure different upper-body, lower-body, and end-effector combinations to suit specific use cases while reducing total cost of ownership.[19]
A distinctive design feature of the HMND 01 is its use of interchangeable protective garments. These coverings serve multiple functions: minimizing contamination risks in food or pharmaceutical logistics, reducing the effects of collisions in environments shared with human workers, and providing a customizable visual appearance. The garments are exchangeable, allowing operators to swap them out for cleaning or to match different work environment requirements.[20]
The HMND 01 is powered by KinetIQ, Humanoid's proprietary AI framework for end-to-end orchestration of humanoid robot fleets. KinetIQ uses a cross-timescale architecture comprising four cognitive layers, each operating simultaneously at different temporal resolutions. Each layer treats the layer below it as a set of tools, orchestrating them through prompting and tool use to achieve goals set by the layer above.[21]
Operating at the seconds timescale, System 3 is an agentic AI layer that treats individual robots as tools in its repertoire and dynamically allocates tasks to optimize fleet-wide operations. It integrates bidirectionally with facility management systems, including warehouse management systems (WMS) and enterprise resource planning (ERP) platforms, to receive task requests, track progress, report completions, and handle exceptions. System 3 is applicable across logistics, retail, manufacturing, and service environments.[22]
Functioning at the second-to-subminute timescale, System 2 uses an omni-modal language model to observe the environment through the robot's sensors and interpret high-level instructions from System 3. It decomposes goals into executable sub-tasks and dynamically updates plans based on real-time visual context rather than relying on fixed sequences. When encountering situations beyond its capability, System 2 can escalate to human operators for intervention.[23]
System 1 is a vision-language-action neural network that commands target poses for subsets of the robot's body parts (hands, torso, pelvis) at a subsecond timescale. It exposes multiple capabilities, including picking, placing, manipulating, and locomoting, which System 2 can invoke as needed. KinetIQ's VLA issues new predictions at 5 to 10 Hz, with each prediction constituting a chunk of higher-frequency actions executed at 30 to 50 Hz by System 0.[24]
The lowest layer of the KinetIQ stack operates at 50 Hz and is responsible for achieving the pose targets set by System 1 while maintaining dynamic stability across all robot joints. System 0 uses reinforcement-learning-trained whole-body control for both bipedal and wheeled robots. This shared control approach enables cross-embodiment capability, meaning a single AI model can control robots with different morphologies, allowing knowledge and skill transfer across the fleet.[25]
Humanoid has established a collaboration with NVIDIA to accelerate robotic capabilities. The partnership involves three key NVIDIA technologies:[26]
| Technology | Role in HMND 01 |
|---|---|
| NVIDIA Jetson Thor | Edge computing processor for on-device execution of robotic foundation models |
| NVIDIA Isaac Sim | Open-source simulation framework for creating digital twins and validating hardware designs |
| NVIDIA Isaac Lab | Open-source learning framework for reinforcement learning locomotion and manipulation training |
Humanoid became one of the first European companies to integrate NVIDIA Jetson Thor into a humanoid robot prototype. The company also integrates NVIDIA's Isaac GR00T N1.7 vision-language-action model to enhance autonomous decision-making. Using NVIDIA's AI infrastructure, Humanoid reports that VLA model post-training can be completed in just a few hours, and a locomotion policy can be trained from scratch and deployed on a physical robot within 24 hours.[27]
In January 2026, Humanoid and Siemens completed a two-week proof-of-concept deployment at the Siemens Electronics Factory in Erlangen, Germany. The trial focused on a tote-to-conveyor destacking task within Siemens' logistics operations, with the HMND 01 Alpha Wheeled robot autonomously removing totes from stacked storage, transporting them across the facility, and placing them onto a conveyor at a designated handover point for human workers.[28]
The deployment achieved the following performance benchmarks:
| Metric | Result |
|---|---|
| Throughput | 60 tote moves per hour |
| Tote sizes handled | 2 different sizes |
| Continuous autonomous operation | 30+ minutes per stretch |
| Daily uptime | 8+ hours |
| Pick-and-place success rate | >90% |
Following the successful trial, both companies indicated that a wider rollout deploying larger numbers of humanoid robots across Siemens facilities could follow, depending on further capability demonstrations.[29]
Over a six-week period in early 2026, Humanoid deployed the HMND 01 Alpha Wheeled at the Ford Innovation Centre in Cologne, Germany. The trial tested the robot in two complex automotive manufacturing workflows: tote handling for kitting operations and dual-arm manipulation of large metal car body parts.[30]
Key results included:
| Metric | Result |
|---|---|
| Pick-and-place throughput | 83 units per hour (target: 50) |
| Continuous uninterrupted operation | 1 hour (target: 30 minutes) |
| Fully autonomous reliability | 97% |
| On-site data collection for model training | 1 hour to generate high-performing autonomous model |
The Ford POC demonstrated the robot's ability to rapidly adapt to new environments with minimal on-site training data. Founder Artem Sokolov commented: "The POC showed that rapid progress is possible when both sides align on scope and maintain commitment to safety."[31]
From January to February 2026, Humanoid completed a proof of concept with SAP and Martur Fompak, a Turkish-headquartered automotive parts supplier, in a live production logistics environment. The trial represented a significant technical milestone: it was the first time the HMND 01 was controlled by an external enterprise system in a production setting.[32]
The HMND 01 Alpha Wheeled received task instructions from the SAP AI agent through the SAP Joule agent layer, autonomously navigated to designated pallet locations, retrieved KLT (Kleinladungstrager) boxes, and delivered them to trolleys. The robot successfully handled three different tote types with an 8 kg dual-arm payload operating limit. SAP Extended Warehouse Management sent tasks to the robot over the internet and managed its actions remotely, without requiring a custom local control system. This allowed the robot to integrate directly into the company's core IT system for orders, inventory, and task management.[33]
The project followed a structured rollout: physical twin development, in-house testing, site preparation, on-site deployment (including setup, training, optimization), and stakeholder demonstrations. Following the POC, the partners announced plans for further on-site validation phases and exploration of more complex use cases within Martur Fompak's production environment.
At NVIDIA GTC in San Jose on March 20, 2026, Humanoid presented a live demonstration of voice-activated multi-robot collaboration powered by KinetIQ. The showcase featured two wheeled HMND 01 robots with grippers operating in a simulated retail environment. Booth visitors issued voice commands requesting the robots to handle items such as water bottles and popcorn. The system interpreted the requests, allocated tasks between the two robots, coordinated physical handovers between them, and returned to default positions while awaiting the next command. Real-time status displays showed task distribution and execution progress.[34]
The HMND 01 enters a rapidly growing humanoid robotics market that has attracted significant investment and attention from major technology companies. Humanoid's industrial focus and modular approach position it alongside several well-funded competitors.
| Robot | Manufacturer | Country | Height | Weight | Target market | Locomotion | Estimated price |
|---|---|---|---|---|---|---|---|
| HMND 01 Alpha Wheeled | Humanoid | UK | 220 cm | 300 kg | Industrial / logistics | Wheeled | $50,000-$70,000 |
| HMND 01 Alpha Bipedal | Humanoid | UK | 179 cm | 90 kg | Industrial / logistics | Bipedal | ~$120,000 |
| Optimus | Tesla | USA | 173 cm | 57 kg | Industrial / consumer | Bipedal | $20,000-$30,000 (target) |
| Figure 02 | Figure AI | USA | 170 cm | 70 kg | Industrial / enterprise | Bipedal | $100,000+ |
| Atlas | Boston Dynamics | USA | 150 cm | 89 kg | Research / industrial | Bipedal | Not for sale |
| NEO | 1X Technologies | Norway/USA | 167 cm | 30 kg | Consumer / home | Bipedal | $20,000 |
| Digit | Agility Robotics | USA | 175 cm | 65 kg | Logistics / warehouse | Bipedal | Not disclosed |
Humanoid differentiates itself from competitors in several ways. Its modular architecture, allowing customers to choose between wheeled and bipedal bases and swap between different end-effectors, provides flexibility that most competitors' fixed-configuration designs do not offer. The shared upper-body design between variants enables skill transfer, meaning tasks learned on the wheeled platform can be deployed on the bipedal version with minimal retraining. The company's RaaS business model lowers adoption barriers compared to the outright purchase required by most competitors.[35]
However, Humanoid faces significant competitive pressures. Tesla plans to manufacture Optimus at scale with a target price of $20,000 to $30,000, leveraging its existing manufacturing infrastructure. Figure AI has raised over $1.5 billion in funding from investors including Jeff Bezos, Microsoft, OpenAI, and NVIDIA. Boston Dynamics brings decades of locomotion research and a production-ready electric Atlas platform with 56 degrees of freedom. Chinese competitors including Unitree, UBTECH, and Agibot are also rapidly advancing their humanoid platforms.
As of early 2026, Humanoid reports the following commercial metrics:[36]
| Metric | Value |
|---|---|
| Pre-orders | 20,500+ |
| Completed proofs of concept | 6 |
| Active pilot programs | 3 |
| Team size | 200+ employees |
| Office locations | London, Boston, Vancouver |
The company's near-term focus is on industrial deployments in logistics, warehousing, and automotive manufacturing. The structured rollout plan begins with establishing the robot's reliability in controlled industrial settings before expanding to service-sector and eventually household applications.