Humanoid Robot Motion Planning Patents: Leaders & Gaps 2026
- Filing peaked in 2025 at 10 records after years of near-zero activity — this is a field whose patent record only just started, not a mature one.
- B25J dominates the IPC mix (19 of 20 records) while G06N (AI models) and G06T (image processing) appear in only a handful — the perception and learning layers are thinly claimed relative to the mechanical/control core.
- The US receives more than half of all filings (12 of 20) with China at 6 and South Korea at 2, so freedom-to-operate work should start with the USPTO register, not a single foreign office.
Filing growth compares 2021 (1 records) with 2024 (3) — 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 20 records in scope (CR5), not by the ranked leaders only.
What this landscape covers
This dataset tracks patent families at the intersection of humanoid, bipedal and legged robot platforms and the specific control problem of whole-body motion planning — footstep planning, trajectory optimization and dynamic motion planning claimed under the manipulator, industrial-control and motor-vehicle-steering IPC classes that this technology has historically been filed under. It is a narrow, control-engineering slice of the broader humanoid robotics field, not a survey of actuators, sensors or general AI.
Twenty published records span 2015 to the 2026 data cut-off, with filing activity concentrated almost entirely in the last three years of that window. Because publication lags filing by roughly 18 months, the 2026 count is necessarily incomplete and the true 2025–2026 filing rate is higher than the raw numbers show.
Filing trend and technology composition
Two views of the same 20-family dataset: how filing activity has moved year over year, and which IPC subclasses carry the claim volume.
Filing trend: a late, sharp ramp
Activity was flat through the mid-2010s and into the early 2020s (2022 sat at just 1 filing), then rose sharply to a peak of 10 in 2025. That shape says the patenting community around humanoid motion planning is very young — most of the technical positions on record were staked in the last two to three years, and the field has not yet had time to consolidate around dominant approaches.
IPC composition: mechanical core, thin AI layer
B25J (manipulators and robots) covers 19 of 20 records — essentially every filing touches this class. B62D (motor-vehicle steering, used here for legged locomotion mechanics) and G05D (control of non-electric variables) follow at 7 and 4. G06N (AI models), G06F (data processing) and G06T (image processing) each appear only once to three times, suggesting that learning-based and vision-based planning approaches are claimed far less densely than the mechanical/control fundamentals.
Shares are the percentage of the 20 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Humanoid Robot Motion Planning with Eureka
This page is one run against one query. Ask Eureka your own question about humanoid robot motion planning and every answer comes back with the patent numbers behind it.
Try EurekaRepresentative filing and most-cited records
Footstep planning method, robot and computer-readable storage medium (US20220040859A1)
A footstep planning method includes: obtaining a number of depth images of an environment in a walking direction of a legged robot; creating a three-dimensional model of the environment based on the depth images; determining at least one even region from the three-dimensional model of the environment; and selecting one or more of the at least one even region as one or more candidate footstep locations for the legged robot to step on.Filed by UBTECH ROBOTICS CORP LTD, published 2022-02-10.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20160243699A1 | Method for developing and controlling a robot to have movements matching an animation character | 31 |
| 2 | KR1020120019893A | Footstep Planning Method for Bipedal Robot | 18 |
| 3 | CN118682750A | 基于在线质心轨迹优化的人形机器人高动态跳跃运动控制方法 | 10 |
| 4 | US20250018560A1 | Learning robust legged robot locomotion with implicit terrain imagination via deep reinforcement learning | 8 |
| 5 | US20260070221A1 | Bipedal action model for humanoid robot | 6 |
| 6 | CN120307288A | 一种人形机器人全身协调行走的运动在线生成与控制方法、系统、计算机可读存储介质及计算机程序产品 | 5 |
| 7 | US20220040859A1 | Footstep planning method, robot and computer-readable storage medium | 3 |
| 8 | US20260097488A1 | Planning and control method for legged robot, apparatus, robot, and storage medium | 2 |
| 9 | US20260126796A1 | Bipedal action model for humanoid robot | 2 |
| 10 | CN120503201A | 考虑滑移转向的轮腿式机器人全身运动控制系统及方法 | 2 |
Citation counts reward older filings simply for having been on record longer; read them as a signal of influence within this corpus, not as a ranking of current technical importance.
Patent titles are shown in the language they were filed in, not translated, so that each record stays verifiable against the original filing — a translated title will not match in Eureka or in any national register. 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 numbers mean for a filing decision
Three read-throughs from the trend, geography and citation data that matter more for strategy than the raw counts alone.
The record is young, not settled
With the midpoint year 2022 at a single filing and the peak arriving only in 2025, most technical positions in this space are recent. There is limited prior art depth to search around for any given sub-approach, and no single filer has had the runway to build a dense thicket yet.
Freedom-to-operate starts at the USPTO
Over half of all publications route through the United States, making it the primary office to clear before commercial deployment there or in export markets that recognize US priority filings. China's six filings and South Korea's two round out a much smaller secondary footprint.
Mechanical and control claims dominate; perception is thin
Nearly every record touches the manipulator/robot control class, but AI-model (G06N), data-processing (G06F) and vision (G06T) classes appear in only a handful of records combined. Claim space around learned perception feeding into footstep or trajectory decisions is comparatively open.
Influence skews toward the oldest filings
The most-cited records in this corpus date from earlier in the window, which is expected — citation counts accumulate with time and favour records that have simply existed longer. Recent high-relevance filings such as the 2025–2026 reinforcement-learning and bipedal-action-model records have not yet had time to accrue citations.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to humanoid robot motion planning, with the prior art for and against each one.
Who is filing, and how concentrated is the field
Twenty families is a small enough set that no single assignee dominates outright, but the momentum data shows several named filers with zero activity in the latest year — a sign that early movers may be pausing or that filings from 2025–2026 have not yet published under their names.
No aggressive continuation filing yet
Every published record in this dataset maps to its own family, meaning no assignee has yet built a multi-continuation thicket around a single core invention. That keeps design-around analysis comparatively simple today, but it will not stay that way if filing activity keeps accelerating.
Filing is mostly solo, with one university partnership
The one identified co-assignee pairing — a Harbin Institute of Technology unit filing jointly with its Suzhou research institute — is the only joint-filing relationship on record. Most other assignees, including corporate and academic filers, hold their families independently.
Several early filers show no recent activity
Assignees including a Korean advanced-institute filer, a Disney entity, and multiple Chinese university and corporate filers show zero filings in the latest year, with some down -100% YoY from a prior single filing. This may reflect genuine pauses, strategic shifts to trade secret, or simply publication lag masking filings still in the pipeline.
| Assignee | Recent year | YoY |
|---|---|---|
| Digital Artificial Intelligence Corporation | 0 | -100% |
| Korea Advanced Institute of Science and Technology (KAIST) | 0 | — |
| Disney Enterprises, Inc. | 0 | — |
| Beijing Institute of Technology | 0 | -100% |
| Suzhou Kean Ke Intelligent Technology Co., Ltd. | 0 | -100% |
| Yale University | 0 | — |
| Tsinghua University | 0 | -100% |
| UBTECH Robotics Corp Ltd | 0 | — |
Where to take this analysis
The dataset flags where claim density sits and where it doesn't; turning that into a filing or freedom-to-operate decision takes a closer read of the specific claims.
Map claim scope on the most-cited records
Start with the highest-cited US and Korean records to understand exactly what locomotion and footstep-planning approaches are already blocked before drafting new claims.
Explore citing and cited records in EurekaTrack the 2025–2026 filing surge as it publishes
Given the 18-month publication lag, several 2025–2026 filings — including reinforcement-learning-based locomotion approaches — are likely still working through the pipeline.
Set up monitoring in EurekaAssess the thin AI and vision claim layers
With G06N, G06F and G06T covering only six of twenty records combined, a deeper technical read of learned-perception approaches to footstep and trajectory planning may reveal genuine white space.
Run a white space analysis in EurekaCommon questions about humanoid robot motion planning patents
This dataset identifies 20 published patent families between 2015 and the 2026 data cut-off, filed under IPC classes covering manipulator/robot control, industrial control systems and motor-vehicle steering mechanics as applied to legged locomotion. That is a small, young corpus compared to more established robotics sub-fields. Because publication lags filing by roughly 18 months, the true number of filings from 2025 and 2026 is almost certainly higher than what has published so far.
The dataset includes filers ranging from Chinese universities and robotics companies to a handful of academic and corporate filers in South Korea and the US, but no single assignee holds a dominant share of the 20 families. Recent-year momentum data shows several named assignees at zero filings in the latest year, which may reflect a shift toward trade secret protection, a pause in filing, or filings still working through the publication pipeline. Reviewing the most-cited records is a useful starting point for identifying which filers' technical approaches have drawn the most follow-on citation.
US20220040859A1, filed by UBTECH Robotics, claims a footstep planning method that builds a three-dimensional model of the walking environment from depth images and selects candidate footstep locations from detected even regions. It does not claim footstep planning broadly — it is specific to depth-image-based terrain modeling and even-region detection as the basis for candidate selection. Approaches using other sensing modalities, learned terrain classifiers, or different region-selection logic sit outside its literal claim scope, though a full freedom-to-operate opinion would need to check dependent claims and file history.
The clearest gap is between the densely claimed mechanical and control core (B25J appears in 19 of 20 records) and the thinly claimed AI and vision layers (G06N, G06F and G06T combined appear in only six records). That imbalance suggests claim space around learned perception feeding directly into footstep or trajectory decisions, terrain-adaptive reinforcement-learning approaches, and multimodal bipedal action models remains comparatively open. Confirming true white space requires checking recent unpublished filings as well, given the 18-month publication lag.
The United States receives the largest share of publications in this dataset at 12 of 20 records, followed by China at 6 and South Korea at 2. That concentration means a freedom-to-operate review should prioritise the USPTO register first, with secondary attention to Chinese and Korean filings depending on where a product will be manufactured or sold. The small number of receiving offices overall also suggests this technology has not yet been filed defensively across a broad set of jurisdictions.
Research Humanoid Robot Motion Planning in depth with Eureka
Go past this page: query the whole humanoid robot motion planning 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.