R-Noid
Last reviewed
May 11, 2026
Sources
9 citations
Review status
Source-backed
Revision
v5 · 2,332 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 11, 2026
Sources
9 citations
Review status
Source-backed
Revision
v5 · 2,332 words
Add missing citations, update stale details, or suggest a clearer explanation.
| R-Noid | |
|---|---|
| General information | |
| Manufacturer | Robot.com (Kiwi Campus, Inc.) |
| Country of origin | United States |
| Founder/CEO | Felipe Chavez |
| Announcement | 2025 (under the new Robot.com brand) |
| Status | In production; deployed |
| Type | Mobile humanoid robot |
| Locomotion | Wheeled (omnidirectional) |
| Height | ~1.6 m (160 cm) |
| Weight | ~50 kg |
| Degrees of freedom | 22 |
| Arm DOF | 7 per arm (dual arms) |
| Torso DOF | 4 |
| Payload per arm | 11 lbs (5 kg) |
| Max speed | ~6 km/h |
| Battery runtime | ~6 hours per charge (reported) |
| Autonomy | Generative VLA (Vision Language Action) |
| Reported price | ~$19,000 (third-party listing) |
| Website | robot.com/r-noid |
R-Noid is a mobile humanoid robot developed by Robot.com, the company legally registered as Kiwi Campus, Inc. The robot is designed for practical deployment on factory floors, in warehouses, and in service venues where it works alongside human staff on repetitive, physically demanding tasks. Examples include semiconductor wafer handling, box moving, kitchen prep, and parts delivery. Robot.com's stated philosophy is "robots for now, not someday," which the company uses to position R-Noid as an immediately deployable platform rather than a research prototype.[1][2][3]
R-Noid sits inside a broader product family of wheeled and quadruped robots that the company launched alongside its 2025 rebrand from Kiwibot to Robot.com. The line includes R-Kiwi for sidewalk delivery, R-Cargo for warehouse transport, R-Dog as a quadruped advertising platform, and the related R-Top stationary humanoid for fixed workstations.[4][5]
The company behind R-Noid was founded in 2016 at UC Berkeley's SkyDeck incubator as Kiwi Campus, by Felipe Chavez, Jason Oviedo, and Sergio Pachon. It originally operated under the consumer brand Kiwibot and built four-wheeled sidewalk delivery robots that completed food orders on college campuses. The first deployment was at UC Berkeley in 2017, followed by Stanford, Sacramento, San Jose, and more than thirty other US universities over the next several years.[5][6]
In May 2025, Kiwibot announced that it had acquired the Robot.com domain and was relaunching under that name. The company framed the rebrand as a shift from being a single-product delivery business to a multi-category robotics firm pursuing the wider industrial automation market. At the time of the announcement, Robot.com reported that its existing fleet had grown from roughly 300,000 cumulative robotic tasks over its first seven years to more than one million tasks completed in the following seven months, and that it operated over 500 robots in more than half of US states.[4][7][8] R-Noid was unveiled as part of the same launch, presented as the firm's first humanoid platform.[7]
| Corporate fact | Detail |
|---|---|
| Legal entity | Kiwi Campus, Inc. (Delaware) |
| Original brand | Kiwibot (2017 to 2025) |
| Current brand | Robot.com (since May 2025) |
| Headquarters | Berkeley, California, United States |
| Founders | Felipe Chavez, Jason Oviedo, Sergio Pachon |
| Year founded | 2016 (incubation at UC Berkeley SkyDeck) |
| Investors (selected) | Headline, UC Berkeley SkyDeck Fund, Sodexo Ventures, New Future Capital, Innosphere Ventures Fund, Tylt Ventures |
| Notable customers | Sodexo, Amazon Web Services, SKIP (Canada), GXO Logistics |
The company's earlier work with delivery robots is relevant because it shaped how R-Noid is positioned. Robot.com leans hard on its track record of fleet operations and remote support infrastructure, rather than on a single technical breakthrough. The pitch is that R-Noid arrives backed by years of running real robots in real environments, not in a lab.[7][8]
R-Noid is a wheeled humanoid with a 22-degree-of-freedom body and a dual-arm upper torso. It stands roughly 1.6 meters (160 cm) tall and weighs about 50 kg, which puts it on the smaller and lighter end of current commercial humanoids. The frame is described as a lightweight aluminum alloy structure with composite shell panels, intended to navigate aisles, counters, and workstations that were designed for people.[2][3]
| Specification | Value |
|---|---|
| Height | ~1.6 m (160 cm) |
| Weight | ~50 kg (with battery) |
| Total degrees of freedom | 22 |
| Arms | Dual 7-DOF |
| Torso | 4-DOF articulated |
| Hand fingers | 5 per hand (Dex-Hand option) |
| Payload per arm | 11 lbs (5 kg) |
| Combined dual-arm payload | Up to ~9 kg (reported third-party figure) |
| Display | Dual-screen expressive face |
| Autonomy stack | Generative Vision-Language-Action model |
| Charging | Autonomous docking |
| Locomotion | Omnidirectional wheeled base |
| Max speed | ~6 km/h |
| Battery runtime | ~6 hours (reported) |
| Connectivity | Wi-Fi, Bluetooth, 5G |
| Operating system | Linux base plus Robot.com service layer |
| Ingress protection | IP32 (reported) |
| Motors | Brushless |
| Gearing | Harmonic drives in arms; planetary gears in joints |
| Camera | 1080p |
| Glass-to-action latency | ~250 to 450 ms (reported) |
Some of these figures, including the 6 km/h top speed, the 6-hour runtime, the IP32 rating, and the ~$19,000 price, come from the third-party humanoid robot directory humanoid.guide rather than from Robot.com's official site, which discloses fewer numbers.[3] Where the official manufacturer page and the directory disagree, the manufacturer page is the more authoritative source; the directory figures are included here for completeness with the caveat that they are externally compiled.
Each arm has 7 degrees of freedom and can lift up to 5 kg (11 lbs) on its own. Both arms together extend that to roughly 9 kg of combined payload according to third-party reporting, which reflects the fact that bimanual handoffs and shared loads are part of the intended use case rather than each arm always operating at its single-arm maximum.[1][3]
The robot uses a modular wrist interface that accepts swappable end-effectors. Robot.com lists parallel grippers and a dexterous five-finger "Dex-Hand" as the main options, which lets a single chassis cover both pick-and-place work and finer manipulation tasks. The torso adds 4 DOF for waist rotation and bending, so the arms can reach lower shelves and surfaces without the base needing to crouch.[1]
R-Noid does not walk on legs. It rolls on an omnidirectional wheeled base, which is a deliberate engineering choice for cost, reliability, and indoor stability. Omni-wheels let the robot strafe sideways, rotate in place, and thread between obstacles without complex gait planning. The platform is built for flat indoor surfaces in factories, kitchens, and warehouses rather than for stairs or rough terrain.[2][3]
The perception stack combines cameras (1080p) with onboard sensors for obstacle avoidance and context awareness. Robot.com positions R-Noid as "safe alongside people," which in practice means real-time obstacle detection, slower travel near humans, and predictable motion profiles. The dual-screen face provides expressive signals (eye contact, gaze direction, simple animations) that act as social cues so that a worker passing by can tell where the robot is about to move.[1][2]
The robot is driven by Robot.com's own software stack, which the company describes as a "generative VLA" architecture. Vision-Language-Action models are a class of foundation models trained to map camera input and natural language instructions to robot actions, and they have become a common architectural choice across the current generation of humanoid platforms, including offerings from Figure AI, 1X Technologies, and Physical Intelligence.[1]
Robot.com has not published a model card, training corpus, or external benchmark results for the R-Noid VLA stack. The runtime is described as a Linux base layer with the company's proprietary service orchestration software on top, and the directory listing notes a glass-to-action latency of roughly 250 to 450 ms for the system.[3] That latency figure is in the range expected for a foundation-model-driven control loop running on embedded hardware with a cloud or edge fallback for heavier inference.
R-Noid targets jobs that are dull, physically repetitive, or staffing-constrained but that still need fine motor control and the ability to work in human-designed spaces. Robot.com's main application categories are listed below.
| Sector | Use cases |
|---|---|
| Manufacturing | Parts delivery, assembly line support, line restocking |
| Semiconductor | Wafer handling, cleanroom material movement |
| Logistics | Material handling, order fulfillment, induction stations |
| Warehousing | Inventory movement, sorting, tote handling |
| Kitchens | Repetitive food prep tasks, station support |
| Retail and service | Greeting, basic guidance, shelf interactions |
Robot.com has highlighted GXO Logistics as one of R-Noid's customer deployments, in roles such as moving semiconductor wafers, boxes, and parts in live warehouse operations.[1] GXO has separately announced multi-year humanoid agreements with Agility Robotics (for the Digit robot), Apptronik (for Apollo), and Reflex Robotics, and a Robot.com pilot would fit GXO's stated strategy of testing several humanoid platforms in parallel rather than committing to a single supplier.[9] The exact scale of any R-Noid deployment at GXO has not been independently confirmed in trade press.
R-Noid is one piece of a wider product line that Robot.com presented at the 2025 relaunch. The other robots cover different form factors and price points so that a single customer can buy delivery, transport, advertising, and humanoid assistance from the same vendor.
| Product | Type | Primary application |
|---|---|---|
| R-Noid | Mobile humanoid | Factory floors, warehouse lines, kitchens |
| R-Top | Stationary humanoid | Precision tabletop work, semiconductor handling |
| R-Cargo | Wheeled delivery robot | Material transport in warehouses |
| R-Dog | Quadruped robot | Patrol, inspection, mobile advertising |
| R-Kiwi | Sidewalk delivery robot | Campus and last-mile food delivery |
The lineup is loosely tiered by mobility and dexterity. R-Kiwi and R-Cargo are wheeled drones that move objects between fixed points. R-Dog adds inspection and a mobile screen for digital out-of-home advertising. R-Top is a stationary dual-arm robot for fixed workstations, and R-Noid is the only ambulatory humanoid with an upper torso designed to operate at the same physical scale as a person.[4][5]
R-Noid is on the small and light end of the current humanoid market. The table below puts its published specs alongside other widely covered platforms; figures for each robot come from manufacturer disclosures and the cited references rather than a single normalized benchmark.
| Robot | Manufacturer | Height | Weight | Locomotion | Notes |
|---|---|---|---|---|---|
| R-Noid | Robot.com (Kiwi Campus) | ~1.6 m | ~50 kg | Wheeled (omnidirectional) | 22 DOF, dual 7-DOF arms, 5 kg per arm[1][3] |
| Digit | Agility Robotics | ~1.75 m | ~64 kg (140 lb) | Bipedal | First commercial RaaS humanoid at scale[9] |
| Apollo | Apptronik | ~1.73 m | ~73 kg (160 lb) | Bipedal | 25 kg (55 lb) payload, ~4-hour battery[9] |
| Figure 02 | Figure AI | ~1.68 m | ~70 kg | Bipedal | Pilots at BMW; vision-language driven |
| Optimus | Tesla | ~1.73 m | ~57 kg | Bipedal | Internal factory pilots |
The practical effect of the wheels-plus-arms design is that R-Noid trades the ability to climb stairs or walk on uneven floors for a simpler, lower-cost platform that is easier to manufacture and to operate for long shifts. That tradeoff is shared by several other vendors aiming at the warehouse and kitchen market segment, where almost all real workspace is flat and most failure modes for a legged robot would be more disruptive than helpful.[3][9]
Several aspects of R-Noid are documented less precisely than for some competing platforms, and any reader weighing the robot against alternatives should be aware of these gaps:
These gaps do not contradict any of the published claims, but they make like-for-like comparisons with larger humanoid programs harder than the public marketing implies.