Robot Perception & Manipulation Patents: Top Companies & Trends 2026
- Filings grew 51% from 2021 to 2024 (39 to 59 records), even as the field's 2019 peak of 84 records has not yet been matched.
- One company alone holds 41 records and the ranked leaders' top 5 combine for 25.3% of all 557 records in scope — concentrated, but with a long tail below.
- Manipulators and control dominate the claim mix B25J covers 66.4% of records, while AI-model classes like G06N sit at just 11.0%, suggesting the perception layer is less crowded than the mechanical one.
Filing growth compares 2021 (39 records) with 2024 (59) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 557 records in scope (CR5), not by the ranked leaders only.
What the robot perception and manipulation patent record shows
Robot perception and manipulation sits at the intersection of mechanical actuation and machine sensing: patents in this set combine claims on end effectors, actuator control and motion trajectory with claims on environment sensing and mapping. The search covers 557 published records filed or published between 2015 and 2026, drawn from families that pair a perception or manipulation term with a supporting technical element such as position estimation or actuator control.
Filing activity peaked in 2019 and has not returned to that level, but the 2021-2024 window shows genuine growth rather than stagnation. Because publication typically lags filing by roughly 18 months, the most recent one to two years in any chart will understate real activity and should not be read as a slowdown.
Filing trends and technology composition
The dataset spans 557 records across manipulators, control systems, imaging and AI-adjacent classes. Class shares are calculated against the full record count and sum to more than 100% because most filings carry several IPC classes.
A growth window inside a longer plateau
Annual filings ran from 44 in 2017 to a peak of 84 in 2019, then eased before climbing again: 39 records in 2021 rose to 59 in 2024, a 51% increase over that three-year span. The 2025-2026 figures are still filling in due to publication lag and should not be read as a decline.
Manipulators dominate; AI and vision classes trail
B25J (manipulators and robots) appears in 66.4% of the 557 records, far ahead of G05B control systems at 14.5% and G06T image processing at 12.6%. AI-model classes (G06N, 11.0%) and non-electric control (G05D, 9.3%) are present but not dominant, and surgical applications under A61B sit at 8.6% — evidence that manipulation hardware and its control loop, not the learning layer, carries the bulk of current claim density.
Shares are the percentage of the 557 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Robot Perception & Manipulation Patent Landscape with Eureka
This page is one run against one query. Ask Eureka your own question about robot perception & manipulation patent landscape and every answer comes back with the patent numbers behind it.
Try EurekaRepresentative and most-cited filings
Real-time control methods and systems for robot manipulation actions based on collaboration between large models and small models
Filed by Tongji University, this application describes a real-time control method that fuses environmental data with instruction text into multi-modal input, encodes it into a feature vector, aligns it via cross-modal token alignment, and uses a large model to train a smaller model for deployment — with pruning applied to reduce runtime cost.Abstract condensed from the original filing for readability.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US5046022A | Tele-autonomous system and method employing time/position synchrony/desynchrony | 321 |
| 2 | US20110071675A1 | Visual perception system and method for a humanoid robot | 300 |
| 3 | US4907169A | Adaptive tracking vision and guidance system | 258 |
| 4 | US20090221928A1 | Motor training with brain plasticity | 242 |
| 5 | US20150190925A1 | Remotely Operating a Mobile Robot | 224 |
| 6 | US20130138246A1 | Management of resources for slam in large environments | 202 |
| 7 | US20170334066A1 | Machine learning methods and apparatus related to predicting motion(s) of object(s) in a robot's environment … | 183 |
| 8 | US20100152899A1 | Systems and methods of coordination control for robot manipulation | 153 |
| 9 | US20170348854A1 | Robotic manipulation methods and systems for executing a domain-specific application in an instrumented envir… | 145 |
| 10 | US20170024877A1 | Methods and Apparatus for Autonomous Robotic Control | 144 |
Citation counts favour older filings that have had more time to accumulate citations within the searched corpus; treat them as a signal of influence, not of current technical relevance.
Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the filing pattern means for strategy
Three signals stand out once the ranking, the class mix and the timeline are read together.
Leadership is concentrated but not locked in
The top 5 assignees combine for 141 of 557 records, 25.3% of the field, while the leader alone holds 41. That leaves nearly three-quarters of filings spread across a long tail of single- and few-filing entrants — real room for a new entrant to build a defensible position rather than compete head-on with an incumbent.
Mechanical manipulation claims dominate the class mix
Two-thirds of records touch B25J manipulator classifications, while AI-model claims (G06N) sit at just 11.0%. That gap suggests the actuation and end-effector layer is more heavily claimed than the perception-and-learning layer that increasingly drives commercial differentiation.
Growth resumed after the 2019 peak cooled
Filings fell back from the 2019 high of 84 before climbing again — 39 records in 2021 to 59 in 2024. Several of the currently top-ranked assignees show no filings in the latest year, which is more consistent with publication lag than with an active pullback from the space.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to robot perception & manipulation patent landscape, with the prior art for and against each one.
| Assignee | Co-assignee | Shared families |
|---|---|---|
| iRobot Corporation | Evolution Robotics, Inc. | 3 |
| DILIGENT ROBOTICS INC | WORSNOP PETER | 2 |
| DILIGENT ROBOTICS INC | THOMAZ ANDREA LOCKERD | 2 |
| DILIGENT ROBOTICS INC | NELLITHIMARU ANJANA | 2 |
| DILIGENT ROBOTICS INC | CHU VIVIAN YAW WEN | 2 |
| NEURALA INC | Trustees of Boston University | 2 |
| Google LLC | ZENG ANDY | 1 |
| DILIGENT ROBOTICS INC | TEJEDA STONE XAVIER | 1 |
Only 10 co-assignee pairs appear in the dataset, and the strongest pairs are repeat collaborations within the same organisation rather than cross-company partnerships — evidence that most work here is filed solo.
Who is filing, and where the gate is set for newcomers
The ranked list spans 100 companies, from an established leader with 41 records down through a long tail of occasional filers. Recent-year momentum figures show several previously active names posting zero filings in the latest year, which fits the expected publication lag rather than a genuine exit.
One company sets the pace, but not by a wide multiple
The top-ranked assignee holds 41 records against a fifth-place figure of 21 and a tenth-place figure of 12 — a gradual taper rather than a cliff, meaning the leader's position is strong but not insurmountable.
The bar to reach the top 10 is modest
With only 12 records needed to sit tenth and 217 records (39.0% of the field) held by the top 10 combined, a focused filer with a coherent claim strategy could plausibly break into the upper ranking within a few years.
Filing here is mostly a solo activity
Co-assignee pairs are limited to 10 across the whole dataset, and the strongest recurring pairs are internal collaborations between an organisation and named inventors rather than joint ventures between separate companies.
| Assignee | Recent year | YoY |
|---|---|---|
| Google LLC | 0 | -100% |
| X Development LLC | 0 | — |
| DILIGENT ROBOTICS INC | 0 | -100% |
| 3M Innovative Properties Co. | 0 | — |
| Kawasaki Heavy Industries, Ltd. | 0 | -100% |
| A Traction Inc. | 0 | — |
| iRobot Corporation | 0 | — |
| NEURALA INC | 0 | — |
Where to take this next
The dataset points to specific follow-up questions depending on whether you are scoping freedom-to-operate, tracking a competitor, or looking for white space.
Check freedom-to-operate against the densest classes
With B25J claims present in 66.4% of records, any new manipulator or end-effector design should be checked against this class specifically before filing, rather than against the field as a whole.
Run a claim clearance searchWatch momentum, not just rank
Several top-ranked assignees show zero filings in the latest year. Track whether that continues once publication lag closes, since it changes which competitor is actually accelerating.
Set up assignee trackingScope the under-claimed branches
Cross-modal instruction alignment and non-electric control show lower density than the core manipulator classes. A first claim drafted specifically in that gap has more room to stand.
Explore white space in EurekaCommon questions about robot perception and manipulation patents
The ranked list covers 100 companies, with a single leader holding 41 records against a field of 557. The top 5 combined account for 25.3% of all records, and the top 10 for 39.0%, which means the field is concentrated at the top but still leaves a long tail of companies with a handful of filings each. This shape is common in robotics: a few large industrial and technology firms anchor the space while specialist and academic filers fill in narrower niches.
Manipulator and robot-arm claims under IPC class B25J appear in 66.4% of the 557 records, making it by far the densest class. Control and regulating systems (G05B, 14.5%) and image data processing (G06T, 12.6%) follow well behind, with AI-model classes (G06N, 11.0%) trailing further still. This suggests that the mechanical actuation and control layer is more heavily claimed than the perception and learning layer, even though the latter is where much current commercial attention sits.
Filings grew 51% between 2021 and 2024, rising from 39 to 59 records, after cooling from a 2019 peak of 84. The apparent drop-off in 2025 and 2026 figures is an artefact of publication lag, since patent publications typically trail filing dates by around 18 months, not a real decline. Reading the most recent one to two years as a slowdown would be a misinterpretation of how patent data becomes visible over time.
Relative to the dominant B25J manipulator class, sub-areas such as cross-modal instruction-to-motion alignment, tactile end-effector feedback, and non-electric variable control (G05D) show meaningfully lower filing density. Surgical applications under A61B (8.6% of records) and recognition-driven grasp planning under G06K (7.5%) also sit below the core classes. These are not empty categories, but they carry less claim density than the mechanical manipulation core, which is where a narrowly drafted first claim has more room to establish a position.
Very limited: the dataset contains only 10 co-assignee pairs across 557 records. The strongest recurring pairs are internal collaborations between an organisation and its named inventors rather than joint filings between separate companies. This indicates that most work in robot perception and manipulation is filed by a single assignee, with cross-company joint patenting still uncommon.
Research Robot Perception & Manipulation Patent Landscape in depth with Eureka
Go past this page: query the whole robot perception & manipulation patent landscape corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.