Robust AI
Last reviewed
Jun 4, 2026
Sources
19 citations
Review status
Source-backed
Revision
v1 ยท 2,322 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 4, 2026
Sources
19 citations
Review status
Source-backed
Revision
v1 ยท 2,322 words
Add missing citations, update stale details, or suggest a clearer explanation.
Robust AI (styled Robust.AI) is an American robotics company founded in 2019 and headquartered in San Carlos, California. It builds collaborative mobile robots and supporting software for warehouses, fulfillment centers, and manufacturing sites, with the stated goal of having machines work alongside human workers rather than replacing them. The company was co-founded by a prominent group of roboticists and AI researchers, most notably Rodney Brooks, the co-founder of iRobot and Rethink Robotics and a co-inventor of the Roomba, and the cognitive scientist Gary Marcus. Its flagship product is Carter, a force-sensitive, camera-guided autonomous mobile robot (which the company markets as a "collaborative mobile robot") that moves goods through a warehouse and can be physically pushed and redirected by the people working near it. Robust AI is best known for an early commercial partnership with the logistics giant DHL Supply Chain.
This article is about the company. For the general machine-learning property of a model maintaining performance under noisy, shifted, or adversarial inputs, see robustness in machine learning.
Robust AI was incorporated in May 2019 in the San Francisco Bay Area. Its founding team combined academic AI researchers with veterans of robotics and large technology companies. The five co-founders were:
| Co-founder | Background | Role at Robust AI |
|---|---|---|
| Gary Marcus | Cognitive scientist; founder of Geometric Intelligence (acquired by Uber in 2016); NYU professor emeritus | Founding CEO (departed early) |
| Rodney Brooks | Co-founder of iRobot and Rethink Robotics; co-inventor of the Roomba; former MIT professor and director of MIT's Computer Science and Artificial Intelligence Laboratory | Chief Technology Officer |
| Anthony Jules | Former COO and VP of product at Redwood Robotics; CTO of Formant; product manager at Google X; a former student of Brooks | COO, later CEO |
| Mohamed R. Amer | Former senior technical manager at SRI International's Center for Vision Technologies, where he led DARPA-funded work on machine common sense and explainable AI | Chief Science Officer |
| Henrik Christensen | Qualcomm Chancellor's Chair of Robot Systems and professor of computer science at the University of California, San Diego; director of UC San Diego's Contextual Robotics Institute | Co-founder and adviser |
At launch the company framed itself less as a robot maker and more as a software company. The original pitch, championed by Marcus, was to build "the world's first industrial-grade cognitive engine for robots": an off-the-shelf software platform that would inject common sense and reasoning into robots, much as a video-game engine gives third-party developers reusable tools. Marcus, a vocal critic of pure deep-learning approaches, argued that robots needed structured world models to operate reliably in messy real-world settings.
Gary Marcus left his day-to-day role as CEO relatively early in the company's life, after which Anthony Jules became chief executive. By the time Robust AI began shipping hardware, Jules was CEO and Brooks was CTO, and the two were consistently described as the company's operating leadership. Marcus, Amer, and Christensen are credited as co-founders but did not run the company through its commercial phase. Marcus went on to become widely known as a public commentator and critic of large language models and the broader generative-AI industry.
Robust AI's plans changed substantially between its 2019 founding and its first product. Rather than selling a horizontal software toolkit to other robotics companies, the company decided to build and sell its own robot. In a July 2022 interview with IEEE Spectrum, Brooks unveiled this pivot and the product that resulted: a collaborative mobile robot roughly the size and shape of a warehouse picking cart. Brooks contrasted the design with conventional autonomous mobile robots, saying most AMRs behave "like goats" that block aisles and force people to walk around them, whereas Robust AI's machine was meant to behave "like a dance partner" that defers to nearby workers. The product was later named Carter.
Robust AI has raised venture funding across several rounds. Reported totals vary by source; the rounds confirmed by primary coverage are summarized below.
| Date | Round | Amount | Lead investor | Notable participants |
|---|---|---|---|---|
| October 2020 | Series A | $15 million | Jazz Venture Partners | Playground Global, Liquid 2 Ventures, Fontinalis Partners, plus angels Jaan Tallinn and Mark Leslie |
| April 2023 | Series A-1 | $20 million | Prime Movers Lab | Future Ventures, Energy Impact Partners, Jazz Venture Partners, Playground Global |
| 2025 | Additional financing | Undisclosed | (existing investors) | APL Ventures, 15th Rock joined the cap table |
The $15 million Series A was announced on 28 October 2020 by then-CEO Gary Marcus and brought total funding to about $22.5 million at that point. The April 2023 round, led by Prime Movers Lab, brought the cumulative total to roughly $35 million and was tied to scaling Carter deployments with pilot customers. In 2025 the company disclosed an additional, unspecified raise alongside an expansion of Carter's capabilities, adding APL Ventures and 15th Rock as investors. Various funding-tracker databases list a higher cumulative figure (on the order of $45 million), but those aggregate totals are not fully corroborated by primary announcements.
Carter is Robust AI's flagship and, as of 2026, only hardware product. It is a wheeled, battery-powered mobile robot designed for material handling inside warehouses and fulfillment centers. Physically it resembles a shelved picking cart, and that is deliberate: the robot is built to slot into existing human workflows and existing buildings rather than to require a purpose-built facility.
Carter navigates using cameras and computer vision rather than lidar. The original Carter carried roughly 16 onboard cameras feeding neural processors; the later Carter Pro uses what the company calls a 360-degree AI-optimized vision system (described in some 2024 coverage as around 20 cameras). Using visual simultaneous localization and mapping (visual SLAM), Carter builds a semantic map of its surroundings: it can recognize and remember the locations of objects such as a person, a truck, or a ladder, rather than treating the world as undifferentiated obstacles. The robot uses an omnidirectional (holonomic) drive, originally built on mecanum wheels, so it can move sideways and rotate in place in tight aisles.
The defining feature is collaboration through touch. Carter has a control stanchion with a handlebar (bright orange on Carter Pro) and uses both active and passive force sensing, so a worker can simply grab it and push, slow, or reposition it by hand, and the robot yields and adjusts rather than fighting the person or triggering a hard safety stop. Carter can also operate autonomously, drive point-to-point on its own, or follow a worker through a facility after a brief setup. Multiple Carters can coordinate as a group, so one person can manage a "train" of several robots instead of pushing one cart at a time. Later versions added a built-in barcode scanner, adaptive LED lighting, and overhead status displays to communicate the robot's intent to people nearby.
Carter is software-defined and multi-functional: the same hardware can be reconfigured between fulfillment (piece) picking, point-to-point transport, and acting as a mobile sorting wall, depending on the workflow a facility needs that day. The robots are coordinated by Robust AI's software suite, Grace, a no-code system that lets warehouse staff map a facility using only a tablet with a camera (no manual joystick driving) and then design and adapt how work is divided between people and robot fleets. Grace dynamically distributes and re-balances tasks across workers and Carters as conditions change.
Robust AI sells Carter primarily through a robot-as-a-service (RaaS) subscription model, in which Grace software is bundled with the hardware, while offering outright-purchase options to larger enterprises. The company positions Carter at third-party logistics providers (3PLs) and other operators that want flexible, scalable automation with a fast return on investment, especially in piece-picking fulfillment.
In October 2024 Robust AI introduced Carter Pro, an upgraded version aimed at commercial deployment, in conjunction with its DHL partnership. Carter Pro emphasized the force-sensitive handlebar, the 360-degree vision system, active and passive force sensing, a built-in barcode scanner, and human-robot-interaction features such as adaptive LED indicators and high-visibility marquees. In 2025 the company further expanded Carter's capabilities so a single fleet could switch among picking, transport, and sorting functions, and it described positive productivity results from the field. Carter's force-sensitive user interface earned Robust AI a 2025 RBR50 Robotics Innovation Award from The Robot Report.
Robust AI's most significant customer relationship is with DHL Supply Chain, the contract-logistics arm of Deutsche Post DHL. On 16 October 2024 the two companies announced a partnership to deploy Carter Pro in DHL warehouses, with initial rollouts in the United States and Mexico. DHL described Robust AI as its fourth strategic robotics partner, and Robust AI characterized the deal as its first commercial agreement. Public coverage of the 2024 announcement framed it as a deployment and co-development partnership; the parties did not disclose an equity investment by DHL or any payment terms, and reports of the unit commitment described it as an initial deployment scaling to "hundreds" of robots over time rather than a fixed large order.
On 2 December 2025 DHL Supply Chain and Robust AI announced a five-year strategic alliance to expand into Latin America, beginning with 15 Carter units in DHL's retail operations in Mexico and a plan to deploy hundreds of Carter robots across DHL's Americas operations in the coming years, with warehouse-management-system integration targeted for 2026. DHL reported that Carter had improved productivity by more than 60% in its North America operations and by roughly 30% in early Mexican operations. As with the 2024 partnership, the publicly released materials describe a deployment and co-development alliance and do not state a dollar amount or an equity stake.
DHL has separately pursued a multi-vendor robotics strategy, pairing Robust AI's collaborative carts with other suppliers such as Locus Robotics for picking and Boston Dynamics for heavy lifting, as part of a broader, multi-hundred-million-dollar automation investment program across its global network.
Beyond DHL, Robust AI has emphasized diversifying its customer base; Brooks said publicly that the company did not intend to depend on a single logistics customer. In November 2025 the warehousing and fulfillment provider Saddle Creek Logistics Services announced that it had deployed Carter collaborative robots to support tote movement and speed up warehouse operations. The company has also reported pilots and deployments with third-party logistics operators in markets including Las Vegas, Nevada, where it said Carter delivered productivity gains on early operations.
Robust AI has drawn attention in the robotics press both for its high-profile founding team and for its contrarian, human-centric design philosophy. Coverage in IEEE Spectrum, TechCrunch, and The Robot Report has highlighted the company's argument that collaboration, force-feedback handling, and camera-based perception make its robots easier to deploy in existing buildings than fully autonomous, lidar-based AMRs that try to remove people from the loop. Skeptics note that the warehouse-automation market is crowded and capital-intensive, with well-funded competitors, and that Robust AI's funding to date is modest relative to some rivals. The company's eventual fortunes are widely seen as tied to whether large logistics operators such as DHL move from pilots to broad, multi-site rollouts.
The company name collides with a common technical phrase. In machine learning, "robust AI" or model robustness refers to a model's ability to keep performing reliably when inputs are noisy, incomplete, drawn from a different distribution than the training data (out-of-distribution generalization), or deliberately manipulated (adversarial robustness). That sense of the term is a property of machine learning systems, not a product or organization, and is unrelated to the warehouse-robotics company described here. The name should also not be confused with Robust Intelligence, a separate AI-security company that was acquired by Cisco in 2024.