Closed-Loop Evaluation Patents: Top Companies & Filing Trends 2026
- Filings peaked in 2022 at 1,617, then declined toward 2026 — the field is consolidating rather than expanding.
- The ranked leader holds 2,539 records, yet the top 10 assignees combined account for only 20.8% of all 38,077 records — no single player controls the space.
- G06F and H04L dominate the classification mix, at 17.8% and 13.3% of records respectively, while image-generation class G06T sits at just 3.0% — a comparatively open branch.
What this patent landscape covers
Closed-loop evaluation of driving policies concerns how autonomous systems are tested against realistic, reactive scenarios rather than fixed, pre-recorded logs. The search underlying this landscape combines closed loop evaluation, driving benchmark and scenario based safety evaluation terms with claim-level language on scenario coverage, reactive agent realism and statistical confidence — the technical vocabulary that separates a genuine closed-loop testing claim from a general simulation or metrics patent.
The dataset spans 38,077 published records filed between 2015 and mid-2026. Because publication typically lags filing by around 18 months, the most recent one or two years in any trend line will read lower than the true filing volume once those applications publish.
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
Filing trends and classification mix
Two views of the same 38,077 records: how filing volume has moved year over year, and which technology classes the claims actually sit in.
A field that has already peaked
Annual filings rose from 914 in 2017 to a peak of 1,617 in 2022, then declined toward 90 in 2026 (a partial year). Read the tail end cautiously — publication lag means 2025 and 2026 figures will rise as pending applications clear, but the shape of the curve past its 2022 midpoint already points to a maturing rather than accelerating field.
Software and networking classes carry the claim density
G06F (electric digital data processing) and H04L (digital information transmission) lead at 17.8% and 13.3% of the 38,077 records respectively, with G06Q, H04W and G06N each in the mid-single digits. Image-focused G06T sits at only 3.0%, and A61B appears at 4.3% — evidence that scenario-generation and perception-adjacent claims are comparatively thin next to the core data-processing and communications classes.
Shares are the percentage of the 38,077 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Closed-Loop Evaluation of Driving Policies with Eureka
This page is one run against one query. Ask Eureka your own question about closed-loop evaluation of driving policies and every answer comes back with the patent numbers behind it.
Try EurekaA representative filing and the most-cited records
US11454949B2 — Auto-correlation of wafer characterization data and generation of composite wafer metrics during semiconductor device fabrication
A controller executes an auto-correlation module that receives patterned wafer geometry metrics and wafer characterization data, determines the correlation between them, ranks the geometry metrics by that correlation, and constructs a composite metric model from the ranked subset.Filed by KLA Corporation, published 2022-09-27. Its claim scope centres on correlation-driven ranking and composite metric construction from multiple measurement sources — a pattern of claim language that recurs across metrics-generation filings in this dataset, well beyond its origin in wafer fabrication.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20120069131A1 | Reality alternate | 2,110 |
| 2 | US20060064411A1 | Search engine using user intent | 915 |
| 3 | US20070156919A1 | Enforcing network service level agreements in a network element | 883 |
| 4 | US20070043860A1 | Virtual systems management | 800 |
| 5 | US20170123487A1 | System and methods for on-body gestural interfaces and projection displays | 738 |
| 6 | US20030128658A1 | Resource allocation for MIMO-OFDM communication systems | 639 |
| 7 | US20140278391A1 | Apparatus and method to classify sound to detect speech | 617 |
| 8 | US5933462A | Soft decision output decoder for decoding convolutionally encoded codewords | 587 |
| 9 | US20060184410A1 | System and method for capture of user actions and use of capture data in business processes | 583 |
| 10 | US6801940B1 | Application performance monitoring expert | 578 |
Citation counts accumulate over time, so older records such as the 2012-published "Reality alternate" filing lead the table by volume of citations rather than by current relevance — treat this as a signal of historical influence, not present-day importance.
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 concentration and citation data imply
Three patterns stand out once the ranking and classification data are read together: no dominant gatekeeper, a plateaued filing curve, and citation weight sitting on older, broadly-worded records.
No single assignee controls the space
The ranked leader holds 2,539 records, but the top 5 combined account for only 15.3% of all records in scope and the top 10 for 20.8%. That leaves the large majority of filings spread across a long tail of single- and few-filing entrants, which means freedom-to-operate analysis has to look well past the household names.
Filing volume has already turned down
Volume rose from 914 filings in 2017 to a peak of 1,617 in 2022 and has since declined toward 90 in the partial 2026 year. Even allowing for publication lag understating the last two years, the field is past its filing peak rather than still accelerating.
Generation and imaging claims are thin
Core data-processing (G06F, 17.8%) and communications (H04L, 13.3%) classes dominate the mix, while image data processing and generation (G06T) sits at just 3.0% of the 38,077 records. That gap is notable given how central scenario generation and rendering are to closed-loop evaluation conceptually.
Citation weight sits on older, broad filings
The most-cited record in this corpus, "Reality alternate" (US20120069131A1), carries 2,110 citations — far ahead of the next closest records. Heavy citation counts on decade-old, broadly worded filings are typical of a mature corpus and say more about historical influence than about which claims matter for a filing decision today.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to closed-loop evaluation of driving policies, with the prior art for and against each one.
Assignee landscape and where activity is slowing
The ranking covers 100 companies counted by patent family. Recent-year momentum figures show that even the most active filers have pulled back sharply from their historical pace.
A large lead that still leaves room below it
The top-ranked assignee holds 2,539 records against a fifth-place figure of 683 and a tenth-place figure of 345 — a steep drop-off that signals a genuine leader rather than a tightly bunched top tier, but one operating inside a field where the top 10 together still hold only 20.8% of all records.
Even top filers are pulling back
The leading assignee dropped to just 3 filings in the latest year, an 85% year-on-year decline, and several other ranked filers show similar or steeper pullbacks, with one dropping to zero. This mirrors the overall filing trend's decline from its 2022 peak.
Co-filing is limited and concentrated
Only 10 co-assignee pairs appear across the dataset, and the strongest of these link a single large filer to individual named inventors or an affiliated subsidiary rather than to independent corporate partners. Cross-company joint filing is not yet a meaningful pattern in this space.
| Assignee | Recent year | YoY |
|---|---|---|
| Qualcomm Inc. | 3 | -85% |
| Cisco Technology, Inc. | 2 | -75% |
| Telefonaktiebolaget LM Ericsson (publ) | 1 | 0% |
| Amazon Technologies, Inc. | 1 | — |
| Google LLC | 1 | -80% |
| International Business Machines Corporation | 0 | -100% |
| Microsoft Technology Licensing, LLC | 0 | -100% |
| Intel Corporation | 0 | -100% |
Where to take this analysis
The figures above answer what has already been filed. The next questions — what a specific claim actually blocks, and where a new filing would sit relative to prior art — need claim-level reading.
Check freedom-to-operate against the top filers
With the top 10 assignees holding only 20.8% of records, a freedom-to-operate review needs to cover the long tail, not just the ranked leaders.
Explore assignees in EurekaRead the claim scope behind the representative filing
US11454949B2's correlation-and-ranking claim language recurs across metrics-generation filings; understanding its actual boundaries takes claim-by-claim reading.
Analyse this patent in EurekaTrack the under-claimed branches before they fill in
Scenario generation and reactive agent modelling carry lower classification share than the field's conceptual centre would suggest — worth monitoring for new entrants.
Set up monitoring in EurekaCommon questions about this landscape
This landscape defines it by claim language covering closed-loop testing, driving benchmarks, and scenario-based safety evaluation, combined with more specific claim-level terms such as scenario coverage, reactive agent realism, and statistical confidence. A patent qualifies when its claims describe evaluating a driving policy against scenarios that react to the system under test, rather than replaying fixed pre-recorded logs. This distinguishes closed-loop evaluation filings from general simulation, mapping, or open-loop metrics patents that use similar vocabulary without the reactive-scenario element.
The ranked leader in this dataset holds 2,539 records, well ahead of the fifth-place assignee at 683 and tenth place at 345. Even so, the top 10 assignees combined account for only 20.8% of the 38,077 records in scope, meaning the majority of filings sit outside the most active companies. A competitive landscape review should therefore look beyond the top-ranked names to the long tail of single- and few-filing entrants.
No — filings rose from 914 in 2017 to a peak of 1,617 in 2022, then declined toward 90 in the partial 2026 year. Because publication typically lags filing by roughly 18 months, the most recent one or two years will always look lower than the eventual total once pending applications publish. Even accounting for that lag, the trend has clearly passed its midpoint peak and is on a downward path rather than an accelerating one.
Classification data shows core data-processing (G06F, 17.8% of records) and communications (H04L, 13.3%) classes dominate filings, while image data processing and generation (G06T) sits at only 3.0%. Given that realistic scenario generation and perception fidelity are conceptually central to closed-loop evaluation, that low share suggests claim space around synthetic scenario generation and reactive agent modelling is comparatively under-claimed. Statistical confidence methods for scenario coverage also appear thin relative to the volume of general metrics claims.
Treat citation counts as a measure of historical influence within the searched corpus, not of present-day relevance. The most-cited record here, a 2012-published filing with 2,110 citations, has had over a decade to accumulate references, which structurally favours older, broadly worded patents over recent, narrowly claimed ones. For assessing current competitive risk, filing recency and claim scope matter more than raw citation count.
Research Closed-Loop Evaluation of Driving Policies in depth with Eureka
Go past this page: query the whole closed-loop evaluation of driving policies 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.