Safe Control and Constraint Satisfaction Patents: Mapping the Field Through 2026
Filing activity peaked in 2020 at 23 records, then settled into a smaller but growing base — 2021's 8 filings rose to 11 by 2024, a +38% increase over that span. Filings rose from zero in 2017 to a peak of 23 in 2020, then dropped back before recovering — 2021's 8 filings climbed to 11 by 2024, a +38% increase over that three-year window. 2025 and 2026 figures are still filling in under the publication lag and should not be read as a slowdown.
- 1MITSUBISHI ELECTRIC CORP10
- 2INTERNATIONAL BUSINESS MACHINE CORPORATION7
- 3HRL LAB7
- 4MITSUBISHI ELECTRIC RESEARCH LABORATORIES INC6
- 5INTER UNIV RES INST RES ORG OF INFORMATION & SYST4
See the full safe control and constraint satisfaction analysis in Eureka
- The complete ranking, not just the top five
- Every IPC branch with its share of the corpus
- The most-cited records, and where claim space is still thin
Common questions about this landscape
What counts as a 'safety filter' or 'recovery policy' in these patents?
In this corpus, a safety filter is a mechanism that intercepts or modifies the output of a learned or optimised control policy before it reaches the actuator, typically to keep the system within a defined constraint set. A recovery policy is the fallback behaviour invoked once a constraint is at risk of being violated, often built on a separate, more conservative controller. Both concepts appear across the G06N and G05B classes in this dataset, sometimes in the same filing where a primary learned policy is paired with a verified backup.
Who are the leading patent filers in safe reinforcement learning and control barrier functions?
The assignee ranking in this landscape covers 59 companies drawn from the full 89 records in scope, with the leader holding 10 records and a steep drop to 4 at fifth place and 3 at tenth. That is a concentrated top tier followed by a long tail of single- or double-digit filers, typical of a field still being actively defined rather than one with an entrenched dominant player. Reviewing the top 10's specific claim scope is more useful than assuming volume alone confers broad protection.
Is patent filing in safe reinforcement learning growing or slowing down?
Filings peaked at 23 in 2020, dropped afterward, then grew again from 8 in 2021 to 11 in 2024 — a +38% increase over that span. The 2025 and 2026 figures in the raw data appear lower, but publication typically lags actual filing by around 18 months, so those most recent years understate real activity rather than showing a genuine slowdown. The 2021-2024 window is the most reliable recent signal, and it points to renewed growth, not decline.
Disclaimer. This analysis is based on Patsnap Eureka data drawn from a limited snapshot of global patent records and is provided for general information and reference only. Patent data carries inherent limitations — recent filings are under-counted because of publication lag, counts may be on a record or family basis, classification and applicant-name data may contain errors or duplicates, and the underlying search query defines the scope shown — so the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing here is an exhaustive prior-art, novelty, freedom-to-operate or validity search, nor does it constitute legal, financial or professional advice, and it should not be relied upon as such. Verify independently and review with qualified patent and legal professionals before acting on it.
Method: Filing trend and technology composition. Derived from a Patsnap search on Safe Control and Constraint Satisfaction covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish. Every share divides by all records in scope. Data: Patsnap Eureka. See the full landscape report.