US 11402467

I'll search for the patent and any related litigation.

Let me search for CAFC/litigation activity and the claim text.

Let me check specifically for litigation and CAFC 2026 activity involving this patent or its owner.

US 11,402,467 — summary

Note on sourcing: The authoritative text I have is the full Google Patents record for US11402467B2 (fetched 2026‑09‑24). My searches of USPTO/PTAB document repositories and general web sources returned no litigation, PTAB proceeding, or CAFC docket naming patent 11,402,467 (the PTAB/petition documents my queries surfaced concerned unrelated patents, e.g. App. 14/656,189, App. 12/380,780, and Patent 7,347,011). I therefore cannot confirm any CAFC 2026 docket for this patent — absence of search hits is not proof none exists, but I found nothing.

Bibliographic data (per the authoritative record)

Discrepancies to flag: A third‑party aggregator (Unified Patents) lists the priority date as 2018‑12‑28, application date 2019‑09‑09, and grant date 2022‑08‑01 — each one day earlier than the Google Patents record. These are most likely time‑zone/date‑convention artifacts (the priority document is a Russian filing dated 29 December 2018 local time). Per the operating rule, I have not "corrected" either set; the Google Patents values are the ones in my authoritative full text.

Abstract (verbatim, condensed)

Methods and computer devices for determining an angular offset of a radar system mounted on a vehicle having a forward direction of travel along a surface. The angular offset is an angle between the scanning direction and the forward direction. The method includes receiving radar data from the radar system; determining projections of an immobile object velocity in the scanning direction and a direction perpendicular to the scanning direction; and determining the angular offset of the radar system based on at least one of the projections of the immobile object velocity.

Plain‑language overview of the independent claims

The specification's "Summary" recites two independent aspects (corresponding to the two independent claims — one method claim and one computer‑device claim). Caveat: the full text supplied to me reproduces the Summary and Detailed Description but not the numbered claim set, so I can state the substance of the independent claims with high confidence but cannot verify their exact claim numbers or precise claim wording.

Independent claim — method. A method of determining an angular offset of a radar system mounted on a vehicle, where the vehicle has a forward direction of travel and the radar has a scanning direction, and the "angular offset" is the angle between the two. The method is performed by a computer device coupled to the radar and comprises three steps:

1. Receive radar data from the radar system, the data containing point‑specific data for a plurality of detected objects, each object's data indicating (i) its position and (ii) its actual radial speed (e.g., Doppler speed).
2. Determine projections of an "immobile object velocity" in (i) the scanning direction and (ii) the direction perpendicular to it. That immobile‑object velocity is associated with a subset of the detected objects that correspond to at least one object immobile with respect to the surface, and the projections are derivable from the actual radial speeds of those subset objects. In practice this is done by RANSAC‑style iterative optimization: fit candidate velocity projections (via Ordinary Least Squares on the per‑object radial‑speed equations), compute estimated radial speeds via Vimob‑cand‑x·(xi/ri) + Vimob‑cand‑y·(yi/ri) = vi‑est, compare to actual radial speeds, discard objects whose error exceeds a threshold, and re‑fit on the reduced set until residuals fall below threshold.
3. Determine the angular offset from at least one of those projections (specifically θ = π − arctan2(Vimob‑y, Vimob‑x)), optionally using the vehicle's own velocity.

Dependent features include: capturing radar data while the vehicle travels strictly forward at substantially constant velocity; positions expressed in spherical (ri, θi) or Cartesian (xi, yi) coordinates; use of RANSAC and OLS; sanity checks that the immobile subset is at least a pre‑determined proportion/number of the detected objects; repetition of the method at a first and later second moment in time; and performing extrinsic calibration of the radar at each such moment based on the corresponding determined angular offset.

Independent claim — computer device. The same subject matter in apparatus form: a computer device (communicatively coupled to the radar) configured to receive the radar data with per‑object position and actual radial speed, determine the projections of the immobile object velocity in the scanning and perpendicular directions for the immobile‑object subset, and determine the angular offset from at least one projection — with the same dependent features (RANSAC, OLS, coordinate conventions, proportion/number thresholds, repeated determination and repeated extrinsic calibration, and iterative reduction of the detected‑object set).

Technical gist

The invention is a self‑calibration technique for automotive radar: while the vehicle drives straight ahead, returns from stationary objects (road signs, parked cars, etc.) all share one velocity vector relative to the vehicle. By robustly fitting that common velocity (RANSAC + least squares over Doppler/range‑rate measurements), the system separates stationary from moving objects and recovers the angle between the radar's boresight/scanning direction and the vehicle's forward axis, which is then used for extrinsic re‑calibration — compensating for radar drift caused by, e.g., bumps, weather, or mounting change, without a dedicated calibration target.

Uncertainty statement

• No CAFC 2026 docket found. Multiple targeted searches produced no federal‑circuit appeal or district‑court action identifying 11,402,467. I state this as "not found," not as "none exists."
• Claim numbering unverified. The independent‑claim overview follows the Summary's two "broad aspects"; the numbered claims themselves are not in the text I hold.
• Assignee/examiner and one‑day date values from third‑party aggregators were not independently confirmed against USPTO Patent Center or the official assignment records; treat the Google Patents values (and the assignment chain shown there) as the best available, and the one‑day offsets as likely time‑zone artifacts.

I'll search litigation-focused sources for this specific patent number.

Let me run more targeted searches to be thorough.

Litigation search: US Patent 11,402,467

Bottom line: I found no litigation — no district-court action, no PTAB proceeding, and no CAFC appeal — naming US 11,402,467. I state this as "not found," not as "none exists."

Sources checked and what they returned

Case-by-case listing

Because no case was found, I cannot provide plaintiff(s), defendant(s), jurisdiction, case number, filing date, or status. There is nothing to tabulate.

Why this is plausible (context, not proof)

The patent's prosecution/litigation posture is consistent with a quiet patent:

• It issued only in 2022-08-02, is a single-inventor, single-family radar-calibration patent (CPC G01S 7/403, G01S 13/931), and its ownership has been shuffled through four corporate hands by assignment rather than by litigation: Yandex.Taxi LLC → Yandex Self Driving Group LLC (2020) → Direct Cursus Technology L.L.C (2023-11-02) → Y.E. Hub Armenia LLC (2024-08-09).
• Current assignee Y.E. Hub Armenia LLC (35 Moskovyan str., Yerevan, Armenia) is the same entity that the Canadian trademark record shows receiving rights from Direct Cursus Technology L.L.C (Dubai) — i.e., a post-2022 holding entity for the former Yandex self-driving IP. The Yandex self-driving assets were spun out as a separate company (Yandex SDG / "Avride" branding) around 2024; I found no assertion campaign tied to this patent.
• The Unified Patents ratings (RUBY "B", CITX "D", SVIX 1.23) indicate a patent not currently flagged as being actively asserted in a campaign.

Explicit uncertainty / limitations

1. I could not run a direct PACER or CourtListener docket query for this patent. My search tooling returns web-indexed documents, not a live PACER "patent number" field search. A case could exist in a non-indexed or recently filed docket.
2. Absence of indexed hits is not evidence of absence. Free/public litigation databases lag filings, and settlement dismissals or sealed complaints may never appear in indexed search results.
3. Verify independently if this matters. The definitive checks are: (a) a PACER "Search by Patent Number" query across all district courts, (b) Docket Navigator's patent-case query, (c) Lex Machina, and (d) the USPTO PTAB Patent Trial and Appeal Board End-to-End (P-TACTS) search for IPR/PGR/CBM petitions targeting 11,402,467.

One cross-reference caveat to the prior section

The earlier summary listed the adjusted expiration as 2040-09-11 (Google Patents). The Unified Patents records surfaced in this round show Expiration Date 2040-09-10, alongside the previously-flagged one-day offsets (priority 2018-12-28, application 2019-09-09, grant 2022-08-01 vs. 2022-08-02). This is the same time-zone/date-convention artifact already noted; I have not "corrected" either value. It does not bear on the litigation question.

PTAB proceedings on file — US 11,402,467

Sourcing note. The canonical proceedings list comes from the structured "PTAB proceedings on file" block in this prompt, which is generated from the USPTO Open Data Portal (ODP) API. That block returns no AIA trial proceedings for US 11,402,467. I ran independent web searches to catch anything ODP has not yet ingested — queries for US 11402467 IPR PTAB, "11,402,467" PTAB inter partes review, Yandex radar patent IPR PTAB "Yandex Self Driving", and "Y.E. Hub Armenia" IPR/PTAB/litigation. None returned a petition, institution decision, FWD, or appeal naming this patent. Per the operating rule, I am reporting this as "not found," not as proof that no petition was ever filed.

Proceedings overview

Total AIA trial proceedings on file: 0 — no IPRs, no PGRs, no CBMs, no derivation proceedings. Breakdown by status is therefore not applicable: zero active, zero claims invalidated, zero claims sustained, zero settled, zero institution denials. The defensive posture this gives a defendant is the opposite of "hardened" and the opposite of "dead": the patent is wholly untested at the Board. All claims that issued on 2022-08-02 stand exactly as granted, and no statutory estoppel under 35 U.S.C. § 315(e) has attached to anyone.

Per-proceeding detail

No proceedings exist to enumerate. I will not manufacture IPR202x-xxxxx numbers to fill this section. For the record, the closest thing to a third‑party "signal" that surfaced is a Unified Patents patent-profile page at https://portal.unifiedpatents.com/patents/patent/US-11402467-B2 — but that is an analytics/ratings page (it displays RUBY, CITX, BRIX, PVIX, RNIX and SVIX scores), not evidence of a Unified Patents-filed challenge. Unified maintains profile pages for enormous numbers of patents it has never challenged. Likewise, the Unified page's date fields (priority 2018-12-28, application 2019-09-09, grant 2022-08-01, expiration 2040-09-10) each sit one day earlier than the authoritative Google Patents record — consistent with the time-zone artifact already flagged in the earlier-generated summary. No inference of PTAB activity should be drawn from those numbers.

Strategic summary

Claim-status picture. There is no IPR-driven narrowing of US 11,402,467. Every claim — the independent method claim and the independent computer-device claim described in the patent's Summary, plus all dependents (the RANSAC/OLS recitations, the Vimob-cand-x·(xi/ri) + Vimob-cand-y·(yi/ri) = vi-est estimation equation, the θ = π − arctan2(Vimob-y, Vimob-x) determination, the pre-determined proportion/number sanity checks, and the repeated extrinsic-calibration-at-two-moments features) — is UNTESTED, not SUSTAINED and not CANCELED. (Caveat carried forward from the earlier section: the full text I hold reproduces the Summary and Detailed Description but not the numbered claim set, so I can state the substance of the claims with high confidence but cannot quote claim numbers from an FWD — because no FWD exists.)

Estoppel landscape. Because no IPR was ever instituted, no § 315(e)(2) estoppel applies to anyone. A defendant today faces a clean slate: it may raise any § 102, § 103, or § 112 ground, in any combination, in any forum, with no "raised or reasonably could have raised" bar. Equally, there is no § 325(e) estoppel from a PGR, and no prior petitioner whose arguments have been tested and rejected. The usual tactical calculus — "is this ground already foreclosed?" — simply does not apply here.

Pattern signals. No petitioner has ever attacked this patent at the Board, so there is no repeat-petitioner pattern and no General Plastic / follow-on-petition risk to weigh. The patent owner (currently Y.E. Hub Armenia LLC, having moved through Yandex.Taxi LLC → Yandex Self Driving Group LLC → Direct Cursus Technology L.L.C → Y.E. Hub Armenia LLC between 2019-09-10 and 2024-08-09) has never had to defend a PTAB appeal, so there is no evidence of aggressive or passive appellate conduct. No defensive aggregator appears anywhere in the chain or in the record. The multi-hop, offshore reassignment chain is itself worth a look for standing/ownership diligence if you are in active litigation, though it has no bearing on PTAB availability.

What the absence means. For a radar/ADAS patent granted in 2022 and held by a well-funded successor entity, zero IPRs by 2026 is a mild negative signal on assertion intensity: heavily asserted patents in the automotive-sensor space (e.g., the LIDAR and radar-calibration families litigated over the past decade) typically attract IPRs within a year or two of assertion. The likeliest explanations are (a) the patent has never been asserted in litigation, or (b) any assertion has been low-profile. I have no litigation record in front of me confirming either — flagging the ambiguity rather than resolving it.

Recommended next steps

1. If you are a defendant being asserted against: you are in an unusually good posture. There is no FWD to link to and no claim to route around — the invention's core is the two-projection RANSAC/OLS fit recited in the independent claims, and the whole claim set is available to challenge. Check your § 315(b) clock: it starts on service of a complaint alleging infringement, and no earlier petition has started it for this patent for anyone else.
2. Strongest-looking IPR theories (subject to a full prior-art search I have not run): the ordinary-least-squares ghost-target/self-velocity estimation literature in automotive radar is deep, and the "fit a common velocity vector to Doppler returns from stationary clutter" technique long predates the 2018-12-29 RU priority. The claim's novelty likely turns on the specific projection equation and the iterative residual-threshold pruning loop — a § 103 attack combining a stationary-object-clutter-fitting reference with a RANSAC-teaching reference is the natural starting point. Treat this as a hypothesis, not a concluded ground.
3. Verify the negative result yourself before relying on it. Pull https://developer.uspto.gov/ptab-api / the ODP trial-proceedings endpoint for application 16/565,886 and patent 11,402,467, and search PTAB E2E at https://ptacts.uspto.gov directly. My conclusion is that no proceeding exists, reinforced by the ODP block in this prompt — but the ODP index can lag, and the consequence of being wrong (filing a duplicative or time-barred petition) is severe.
4. No CAFC appeal to track. Consistent with the earlier-generated summary, no Federal Circuit docket naming 11,402,467 was found in 2026 or any other year. There is no Case: 2x-xxxx to link and no CourtListener record to cite; I am not supplying a docket number I cannot source.

Explicit uncertainty. I found zero PTAB proceedings and zero appeals. I state that as absence of evidence in the sources available to me, not as an affirmative certification that none exists. The single most reliable next action is a direct PTAB E2E / ODP lookup before you build any defense strategy on this section.

I'll research the assignment chain and the entities involved.

Let me dig into the specific assignment records and correspondents for this patent family.

Ownership & Assignment Reconstruction — US 11,402,467 B2

Sourcing note (read first). I was unable to open the USPTO Assignment Center record for this patent in this session; a direct query of assignmentcenter.uspto.gov / assignment.uspto.gov was not retrievable, and searches keyed to application 16/565,886 and to the successor entities returned no reel/frame numbers and no correspondent-of-record on this patent's own assignment documents. Accordingly:

• The chain structure and dates below come from the Google Patents legal-events record (authoritative full text supplied, fetched 2026‑09‑24) plus corroborating foreign-registry records.
• Reel/frame values and the printing correspondent on each of this patent's four recordings are UNKNOWN to me. Per your constraints, I have not invented them.
• Where I cite a correspondent, it is from a sibling Yandex recording whose cover sheet I did retrieve (legacy-assignments.uspto.gov PDFs) — that is an inference by family, not a verified entry on this patent. Flagged as such throughout.

Inventors

Pattern observations: there is only one inventor, so the "all inventors depart within 12 months" fire-sale precursor is not measurable here (n=1). I found no evidence of Sumin leaving the Yandex organisation, and no second US patent naming him surfaced as an assignee-defection signal. The employer-side pattern is the more notable one: the invention was captured by an employee-for-hire act assigned to the operating subsidiary (Yandex.Taxi LLC) at filing, then routed through two group holding entities — normal for a corporate group, not an inventor-driven transfer.

⚠️ Flag on the Russian counterpart: RU2742323 is held by Yandex Self Driving Technologies LLC, not by the US-chain entities. The US and RU chain-of-title are therefore not mirror images; do not assume the Dubai/Armenia transfers reach the RU member.

Original assignee

• Named on the issued patent (Google Patents "Original Assignee"): Yandex Self Driving Group LLC.
• Assignee at filing (2019‑09‑10 event): Yandex.Taxi LLC — the two differ because a group re-assignment (Yandex.Taxi LLC → Yandex Self Driving Group LLC) was recorded 2020‑07‑28.
• Primary line of business: the Yandex autonomous-driving vehicle programme — self-driving cars and autonomous delivery robots; the patent itself is a production-grade automotive radar self-calibration method, and Yandex/Yandex-affiliated entities publicly ship self-driving vehicles and delivery robots (e.g. Y.E. Hub Armenia is the assignee on later robotic-vehicle and LIDAR grants such as US 12,252,921 and US 12,276,732).
• Product embodying the claims: yes, plausibly — the claimed method is an in-vehicle radar boresight recalibration routine; it is the kind of code that runs on the vehicle, not a paper asset. I have no independent engineering confirmation that this specific routine shipped.
• Current status: the original assignee no longer holds the patent. The Yandex group went through a corporate split/divestiture in 2023–2024; contemporaneous Armenian/regional press (RB.RU, Telesputnik — both surfaced in this session) describes Direct Cursus Technology L.L.C (Dubai) as a group-linked holding of international brands (Playhop, Yango, YDB) and ООО «Я.E. Hub Армения» / Y.E. Hub Armenia LLC (Yerevan, 35 Moskovyan str.) as a Yandex-affiliated office; the same reporting notes Yandex's public affiliate lists do not name Direct Cursus. Treat "Direct Cursus is Yandex-affiliated" as strongly evidenced but not formally admitted by Yandex.

There is no bankruptcy of the original assignee that I can verify; the movement of assets is a group reorganisation, not a Chapter 7/11 sale.

Assignment timeline

⚠️ Reel/frame: NOT RETRIEVED for any of the four links below. Dates are from the Google Patents legal-events record. The recorder/correspondent on this patent's own cover sheets could not be confirmed.

1. 2019‑09‑10 (executed; same day as filing) / recorded 2019‑09‑10 — Reel NNNNNN/NNNN (not retrieved)

• Conveyance: Assignment (inventor-to-company)
• Assignor: Anton Aleksandrovich Sumin (sole inventor)
• Assignee: Yandex.Taxi LLC
• Correspondent: not verified on this patent. Family-level evidence: Yandex's inventor→company recordings of this vintage were filed by BCF LLP (1100/2500 Boul. René-Lévesque O, Montréal, QC), submitters Olga Pavlyuk (docket 40703‑119, reel 051572/0285 — Yandex.Technologies LLC, recorded 2020‑01‑21) and Marcela Bondareva (docket 40703‑162, reel 052572/0109 — Yandex.Technologies LLC, recorded 2020‑05‑05). Both are the same firm, same docket series, same Yandex employer-side capture programme.
• Context: acquisition — statutory employee-invention capture, bundled with the US filing.

2. 2020‑07‑28 (recorded) — Reel NNNNNN/NNNN (not retrieved)

• Conveyance: Assignment / internal re-assignment
• Assignor: Yandex.Taxi LLC
• Assignee: Yandex Self Driving Group LLC
• Correspondent: not verified (see BCF LLP note above).
• Context: internal reorganisation — the patent is moved out of the taxi operating entity into the dedicated self-driving group company.

3. 2023‑11‑02 (recorded) — Reel NNNNNN/NNNN (not retrieved)

• Conveyance: Assignment
• Assignor: Yandex Self Driving Group LLC
• Assignee: Direct Cursus Technology L.L.C (Al Barsha First, Al Khaimah Building II, Office No. 3F‑78, Dubai, UAE)
• Correspondent: not verified for the US recording. Cross-jurisdiction corroboration: the UKIPO design record 80814350006000 shows assignment DRC000032137 — Yandex Europe AG → Direct Cursus Technology L.L.C, effective 2023‑09‑12, received 2023‑10‑16, actioned 2023‑11‑13 — and a change of representative from BCF LLP (Jonathan D. Cutler) to Keltie LLP recorded 2023‑10‑03. So the Yandex→Direct Cursus asset migration was a coordinated multi-jurisdiction programme in Sep–Nov 2023, handled by the same Montreal firm at the transition point.
• Context: transfer out of the Russian-origin operating group into the Dubai international holding entity — i.e. a de-Russification / asset-migration step, executed ~2 months before the US recording.

4. 2024‑08‑09 (recorded) — Reel NNNNNN/NNNN (not retrieved)

• Conveyance: Assignment
• Assignor: Direct Cursus Technology L.L.C
• Assignee: Y.E. Hub Armenia LLC (35 Moskovyan str., 0002 Yerevan, Armenia)
• Correspondent: not verified for the US recording. Cross-jurisdiction corroboration: CIPO trademark application 2,333,690 records the same pair with date of change 2024‑10‑09 (registered 2024‑11‑27), Direct Cursus → Y.E. Hub Armenia LLC.
• Context: internal reorganisation — second hopper of the same migration, out of the Dubai entity into the Armenian operating affiliate. 10 days under 24 months after link 3.

No other recordings found: no security agreement, no license, no merger conveyance, no release, no correction, and — critically — no post-2024 assignment. Y.E. Hub Armenia LLC appears to be the terminal holder.

Discrepancy flagged (carried forward from the prior section): Unified Patents' record shows priority 2018‑12‑28, filing 2019‑09‑09, grant 2022‑08‑01, expiry 2040‑09‑10 — each one day earlier than Google Patents (2018‑12‑29 / 2019‑09‑10 / 2022‑08‑02 / 2040‑09‑11). I have not "corrected" either set; the one-day offsets are consistent with local-vs-UTC date stamping.

Timeline diagram

timeline
    title Ownership of US 11402467
    2018 : Priority RU application filed
    2019 : Filed in US by Yandex.Taxi LLC
         : Inventor assigns to Yandex.Taxi LLC
    2020 : Moved to Yandex Self Driving Group LLC
    2022 : US patent granted
    2023 : Assigned to Direct Cursus Technology LLC
    2024 : Assigned to Y.E. Hub Armenia LLC

NPE / troll-pattern signals

Verdict

Operating-company assertion — selected as the nearest of the five permitted labels, with a caveat the taxonomy does not accommodate.

Justification: the chain runs Yandex.Taxi LLC (2019‑09‑10) → Yandex Self Driving Group LLC (2020‑07‑28) → Direct Cursus Technology L.L.C (2023‑11‑02) → Y.E. Hub Armenia LLC (2024‑08‑09) — every party an affiliate of the same self-driving operating group, each hop corroborated by an independent registry (UKIPO DRC000032137, effective 2023‑09‑12; CIPO 2,333,690, change dated 2024‑10‑09), and terminating in a Yandex-affiliated entity holding a 39-patent, 21-CPC-class product portfolio (robotic vehicles, LIDAR, ML). No signal of NPE behaviour is present: no assertion, no defensive aggregator, no asserter-of-record, no bankruptcy sale.

Caveat you should record: the "assertion" limb is not evidenced — I found no suit naming this patent, so the patent is, on the observable record, a non-asserting operating-company asset, which none of your five verdicts describes exactly. Also note the verification gap driving the conservative posture: reel/frame and correspondent-of-record for all four US recordings were not retrieved, so signal #3 could not be conclusively scored.

Verify directly at: https://assignmentcenter.uspto.gov/ (legacy: https://assignment.uspto.gov/patent/index.html) — search patent number 11402467 and application 16/565,886; request the four cover sheets and record the reel/frame and correspondent for each link.

I'll search for the patent's citation records and any USPTO data for US 11,402,467.

Let me pull the specific citation tables for US 11,402,467.

Let me try to retrieve the patent's own (56) References Cited list.

Prior‑Art Analysis — US 11,402,467 (application 16/565,886)

0. Identification and a necessary scoping note

I found no results that actually belong to a different patent number, so the ID resolves cleanly. Before the citations, one honesty requirement:

Scope limitation: The authoritative full text I hold for US 11,402,467 reproduces the abstract, Summary and Detailed Description but truncates before the front‑page "(56) References Cited" list and the numbered claims. Multiple targeted searches (patentimages PDF path, "References Cited" queries, Patent Public Search/PatFT) did not surface the front page of US 11,402,467 itself. I therefore cannot reproduce the patent's own (56) list verbatim, and I will not invent one. What follows is (a) what I could verify from the record, and (b) the best available proxy — the search‑report citation list of the identical‑invention RU family member RU 2742323 C1, plus art surfaced in the citation graph.

Third‑party aggregator Unified Patents records "Patent Art (41)" and "Non‑Patent Literature (2)" for this patent and "Referenced By (0)". I could not open the individual 41 entries, so that count is reported as a figure, not as an enumerated list.

1. What is verified about the citation record

1.1 Counterpart search‑report citations (RU 2742323 C1, same invention)

The FIPS record for RU 2742323 C1 states, in the (56) field («Список документов, цитированных в отчете о поиске» — documents cited in the search report):

US 20180190046 A1, 05.07.2018. US 8125372 B2, 28.02.2012. EP 1436640 A2, 14.07.2004. JP 2015187589 A, 29.10.2015. US 20170108863 A1, 20.04.2017. DE 102015016888 A1, 27.10.2016. US 20180178722 A1, 28.06.2018. EP 2073034 A2, 24.06.2009.

All eight predate the 2018‑12‑29 effective filing date, so all are facially eligible under AIA 35 U.S.C. § 102(a)(1)/(a)(2) (the application is post‑AIA; priority 2018‑12‑29, filed 2019‑09‑10). Note two of them are within ~6 months of priority (#1 and #7) — that does not matter for §102(a)(1) (publication before the effective filing date suffices) but is worth flagging for any §102(b) grace‑period/derivation argument.

Caveat I must state plainly: the FIPS excerpt gives the list but not the X/Y/A relevance categories assigned by the RU examiner. I also could not retrieve the technical content of references #1–#8 in these searches, so I will not fabricate what each discloses. My descriptions below are labelled by confidence.

1.2 Forward citations observed (NOT prior art — context only)

These appeared in Google Patents cross‑reference tables as listing US 11,402,467 / US 2020/0209354 in a "Cited By" position:

• EP 3415943 A1 — Error estimation for a vehicle environment detection system (filed 2017-06-13, pub. 2017-06-13) — "Cited By (3)" includes US 11,402,467 B2.
• US 2018/0120414 A1 / US 11,237,248 B2 — Automated vehicle radar system with self‑calibration (priority 2016-10-31).
• US 2015/0070207 A1 — Method and apparatus for self calibration of a vehicle radar system (appears in the same similar‑art cluster).

Direction of these relationships is ambiguous from the search snippets (a "Cited By" table on a family page can reflect either direction), so I flag them as leads, not as confirmed §102 references.

1.3 Highly relevant art surfaced by research (citation status to US 11,402,467 unconfirmed)

• US 10,088,553 B2 — Pose estimation for vehicle radar (granted 2018-10-02). Claims recite identifying static objects, and calculating an azimuth orientation using a range‑rate equation "defining the range rate of each static object in terms of a longitudinal component and a lateral component, writing the equation in a form which isolates the sine and cosine of the azimuth orientation … and solving for the unknown vector in a quadratic constrained least squares calculation using static object measurements over a period of time."
• US 10,656,246 B2 — Method for estimating a vehicle radar system misalignment using a stationary object detected a plurality of times, correction factors, and minimisation of an error/cost value.
• US 9,903,945 B2 — radar‑based vehicle motion estimation using verified stationary objects and equations for horizontal/vertical velocity components (Vx, Vy) from range‑rate and azimuth data.

These are the closest technical neighbours to the claimed subject matter that I located, but I have not confirmed that any of them is on the (56) list of US 11,402,467.

2. § 102 anticipation analysis

Claim framework used. Per the Summary (and per the RU counterpart's claim 1, which I did retrieve in full), the independent subject matter is:

• Claim 1 (method) — receive radar data with per‑object (position, actual radial speed); determine projections of an immobile object velocity in (a) the scanning direction and (b) the perpendicular direction, the immobile‑object velocity associated with a subset of detected objects immobile w.r.t. the surface and derived from their actual radial speeds; determine the angular offset from at least one of those projections.
• Computer‑device independent claim — same limitations in apparatus form.
• Dependent features: candidate projections + estimated radial speeds + threshold test (RU claim 2/US claim 2‑ish); iterative removal and re‑fit (RU claim 3); the *vi‑est = Vimob‑cand‑x·(xi/ri) + Vimob‑cand‑y·(yi/ri)> equation; θ = π − arctan2(Vimob‑y, Vimob‑x); RANSAC; OLS; spherical vs. Cartesian coordinates; Doppler speed; proportion/number sufficiency checks; repetition at first/second time and extrinsic calibration each time.

Confidence key: ● verified content · ◐ partial knowledge · ○ content not verified.

2.1 Where anticipation is unlikely on the current record

• The RANSAC + iteratively‑reduced‑set dependent claims. Converting the estimator into an iterative consensus algorithm that removes objects exceeding a residual threshold and re‑fits on the reduced set is a specific combination. The RU‑cited references #1–#8 would have to disclose the iterative‑reduction loop itself (not merely robust fitting) to anticipate. Without their full text I rate this unassessed, not "anticipated."
• The θ = π − arctan2(Vimob‑y, Vimob‑x) dependent claim. Very specific formula; anticipation would require an identical or mathematically equivalent expression.
• Repeated determination + extrinsic re‑calibration at first and second moments. Anticipation requires a per‑cycle recalibration loop, which is a narrower teaching than one‑time misalignment estimation.

2.2 Practical conclusion

On the evidence I could obtain, the broadest independent claims (method claim 1 and the computer‑device claim) are the exposed ones, and the closest art is the static‑object/range‑rate pose‑estimation family (notably US 10,088,553 B2 and US 9,903,945 B2) rather than the eight RU search‑report entries — but I cannot confirm which references are actually on the US (56) list, and I could not read the disclosures of references #1–#8. Any firm §102 mapping requires pulling (i) the actual front page of US 11,402,467 and (ii) the full texts of the eight RU‑cited documents.

3. Contradictions / discrepancies to flag

1. Date offsets (carried over and consistent with the earlier section). Unified Patents lists priority 2018‑12‑28, application 2019‑09‑09, grant 2022‑08‑01 — each one day earlier than the Google Patents record (2018‑12‑29 / 2019‑09‑10 / 2022‑08‑02). The RU priority document itself is dated 29.12.2018, supporting the Google Patents values; treat the Unified figures as likely time‑zone artifacts. Not auto‑corrected.
2. Unified Patents says "Referenced By (0)" while Google Patents pages for EP 3415943 A1 and US 2018/0120414 A1 show US 11,402,467 in citation tables. These are different (forward‑ vs back‑citation) counts on different records; I flag the inconsistency rather than reconcile it.
3. "41 patent art + 2 NPL" (Unified) is a count I could not open item‑by‑item, and it is larger than the 8 references in the RU counterpart search report — so the US record very likely contains citations beyond the RU list. My table should not be read as the complete (56) list.

Bottom line: I can confirm the patent identity and the eight citation references carried by its Russian family counterpart, and I can name the technically closest art; I cannot honestly reproduce the US front‑page (56) list or claim‑by‑claim §102 rejections, because the (56) list and the target references' full texts were not retrievable in this session.

I'll note at the outset that the sections provided do not contain a "Prior Art" section — only the patent summary and litigation summary were carried forward. Since the task directs me to use a prior-art section that isn't present, I will (a) flag that gap explicitly, and (b) build the § 103 analysis on prior art I can locate and verify through search, citing each reference. Let me find the most relevant references.

Strong results. Let me pin down the key references' identifiers, dates, and disclosures.

Let me retrieve the full granted claim set to anchor the analysis precisely.

Prefatory note on the assigned source

The task instructs me to "use the results from the Prior Art section of this page." No Prior Art section was supplied in the material carried into this prompt — only the Patent Summary and the Litigation Summary were provided, and the Google Patents full text I hold contains neither a "Citations" nor a "Cited By" table. I therefore flag this discrepancy explicitly, and I have built the § 103 analysis on prior art I located and verified by search, with identifiers and URLs cited. If the intended Prior Art section listed different references, this analysis must be re-run against that list.

Two further cross-reference flags against the previously generated sections:

• The earlier summary correctly noted the claim set was unavailable. I have since located a partial rendering of the granted claim text (Justia). The excerpt of claim 1 includes an iterative-process limitation and a closing step "performing … calibration of the radar system based on the angular offset" — the word used is calibration, not the extrinsic calibration language of the Summary. I treat this as the best available claim text but flag it as unverified against the printed patent.
• The earlier sections' one-day date discrepancies (time-zone artifacts) are immaterial here and are not repeated.

Obviousness analysis — US 11,402,467 B2 under 35 U.S.C. § 103

1. Governing framework and critical date

Because the application has an effective filing date of 2018‑12‑29, two classes of prior art are available:

• § 102(a)(1) art — publicly available (patented, described in a printed publication, in public use, on sale) before 2018‑12‑29.
• § 102(a)(2) art — U.S. patents and U.S. application publications that were effectively filed before 2018‑12‑29 and name another inventor, even if they published later (secret prior art).

The level of ordinary skill is that of a knowledgeable artisan with, e.g., a bachelor's degree in electrical/computer engineering and ~2–3 years of automotive radar signal-processing experience (or a master's degree and ~1 year), familiar with Doppler/range-rate processing, FMCW radar, RANSAC, and least-squares estimation. RANSAC itself is long-standing prior art (Fischler & Bolles, Random Sample Consensus, Communications of the ACM, vol. 24, no. 6, pp. 381–395, 1981).

One priority caveat that changes the art set. If the Russian priority claim were not perfected (e.g., insufficient support for the iterative RANSAC claim scope), the effective date moves to 2019‑09‑10, and additional references with effective filing dates between 2018‑12‑29 and 2019‑09‑10 become available under § 102(a)(2) — notably US 2020/0400814 A1 (Zenuity AB; priority 2019‑06‑17; auto-alignment of radar sensors using stationary-target screening plus a linearized model in alignment angles, longitudinal/lateral velocity and yaw rate, then offset compensation). I treat that reference as a contingent addition.

2. Prior art identified

3. Element-by-element mapping of independent claim 1

Claim 1 (as excerpted, flagged above) requires, at a given moment during radar operation: (a) receiving radar data with per-object position and actual radial speed; (b) determining candidate projections of velocity in the scanning direction and perpendicular direction; (c) computing estimated radial speeds from those candidates and the object positions; (d) when the residual for an object exceeds a threshold, removing it and, in a following iteration, re-fitting on the reduced set; (e) when residuals fall below threshold, taking the reduced set as the immobile subset and the fitted candidates as the immobile object velocity projections; (f) determining the angular offset from at least one projection; and (g) performing calibration based on that offset.

Result: PA‑1 alone discloses elements (a)–(e) and the RANSAC/least-squares sub-features of claims 9–11 with striking specificity — the same per-detection linear radial-speed model in x_i/r_i, y_i/r_i form, solved by least squares, with RANSAC separating immobile from moving objects. PA‑2/PA‑3 supply elements (f) and (g).

4. Why a PHOSITA would have combined these references

The KSR / Graham motivations are unusually strong here; this is not a case of disparate arts stitched together with hindsight.

1. Same field, same problem, same data. All of PA‑1 through PA‑7 are automotive-radar calibration/ego-motion references operating on the identical input: per-detection angle + Doppler/range-rate from radar returns of stationary objects. Ford's stated motivation (accurate ego-motion for autonomous vehicles, degraded wheel-odometry/IMU) and the misalignment references' motivation (boresight drift from manufacturing tolerance, bumps, weather, impacts) are the same engineering problem the '467 patent addresses, and the '467 specification itself recites those same causes of drift (bumps, weather).

2. Ford expressly parameterizes its model by the mounting angle. PA‑1's disclosure states that each radar's stored installation parameters include the mounting angle and the radar's position on the vehicle, and that these enter the linear model. An artisan working from PA‑1 is therefore directly handed the parameter that the '467 patent estimates. The only delta is estimating/verifying that angle rather than treating it as known — precisely the teaching of PA‑2, PA‑3 and PA‑4.

3. The final computational step is a one-line consequence. Once RANSAC + least squares yields the immobile-object velocity components (V_imob‑x, V_imob‑y) in the radar's own frame, the angle between that velocity vector and the vehicle's forward direction is the boresight offset. The '467 claim 3 formula θ = π − arctan2(V_imob‑y, V_imob‑x) is nothing more than applying a two-argument arctangent to the already-estimated vector. PA‑3 (GM '955) already computes a sensor misalignment angle with an arctan2-type expression from stationary-object velocity data; substituting Vy, Vx for the GM matrix elements is the use of a known mathematical tool for its known purpose.

4. Thresholded, iterative outlier rejection is the canonical RANSAC algorithm. The '467 claim 1 loop (fit on the current set → compute residuals → drop objects above threshold → re-fit on the reduced set → accept when residuals are below threshold) is the classic RANSAC formulation of Fischler & Bolles (1981), which PA‑1 expressly implements ("performing … a random sample consensus (RANSAC) calculation to select the measurement data of the stationary objects and disregard the measurement data of the moving objects"). Adding an explicit residual threshold and per-iteration pruning is a routine implementation choice, not an inventive departure.

5. Threshold/proportion sanity checks are taught. '467 claims 12–15 (immobile subset must be at least a pre‑determined proportion or a pre‑determined number of detections, with verification) are met almost verbatim by PA‑2, which compares the percentage fraction of stationary detections to a threshold value, and by PA‑2's and PA‑4's considerations of detection counts and angular cut-values.

6. Predictable results, finite solution set (KSR). Combining (i) RANSAC + least-squares stationary-target fitting, (ii) a threshold on Doppler-vs-predicted-radial-speed residuals, and (iii) an arctan2 based angle extraction yields the predictable, expected result — a robust estimate of the radar's angular offset — with no change in the principle of operation of any reference. Each reference performs exactly its own known function in the combination.

7. Contemporaneous art shows the combination was the obvious route. The 2013–2019 literature and patent landscape (e.g., Choi et al., "Automatic radar horizontal alignment scheme using stationary target on public road," EuMC 2013; Kellner et al., "Joint radar alignment and odometry calibration," Fusion 2015; Ikram & Ahmad, "Automated radar mount-angle calibration in automotive applications," IEEE RadarConf 2019) shows that joint estimation of radar alignment and vehicle motion from stationary returns was an established, actively pursued approach by the priority date.

5. Claim-by-claim conclusion

Bottom line: claims 1–15 are vulnerable under § 103 on the primary combination PA‑1 + PA‑2 + PA‑3 (Ford '777 + EP 3 239 737 + GM '955), with Fischler & Bolles as corroborating NPL for the iterative RANSAC structure and PA‑4/PA‑5/PA‑6/PA‑7 as cumulative art. The strongest single-reference attack is on the sub-features of claims 2, 8, 10 and 11 (all squarely disclosed by PA‑1); the strongest two-reference attack is on claim 1 itself.

6. Counterarguments the patent owner may raise, and responses

• "Ford's RANSAC is not used for a threshold-based per-object pruning loop." Response: RANSAC is definitionally an iterative fit-with-inlier/outlier-rejection algorithm (Fischler & Bolles 1981); Ford's stated goal is precisely to "disregard the measurement data of the moving objects." Expressing the rejection as a residual threshold and re-fitting is implementation detail.
• "Ford solves for ego-velocity, not for the radar mounting angle." Response: Ford's model expressly contains the mounting angle as a stored installation parameter, and the objective outcome — the velocity vector in the radar frame — differs from the '467 output only by the elementary arctan2 step taught by PA‑3.
• "None of the references discloses the exact formula of claim 3." Response: The formula is a mathematical relationship that follows inevitably from the physical geometry once (V_imob‑x, V_imob‑y) is known; a claim reciting only the exploitation of a mathematical relationship for its conventional result adds no patentable weight to the obvious determination (cf. Parker v. Flook; In re Grams). The same reasoning further exposes claims 2, 3 and 11 to a § 101 challenge, which I note but do not develop here (the task is limited to § 103).
• "Secondary considerations." Response: I found no evidence of unexpected results, long-felt-but-unsolved need, industry praise, licensing-due-to-the-invention, or copying attributable to this patent. The ownership transfers in the file are corporate asset reorganizations (Yandex.Taxi LLC → Yandex Self Driving Group LLC → Direct Cursus Technology L.L.C → Y.E. Hub Armenia LLC), not evidence of nexus-bearing commercial success. Absent such evidence, the Graham secondary factors do not rebut the prima facie case. (If such evidence exists, it would need to be gathered and its nexus to the claimed iterative-threshold feature assessed.)

7. Explicit limitations of this analysis

1. No Prior Art section was provided. The instruction to rely on a "Prior Art section of this page" could not be executed literally; the references above are my own search results. If the intended section listed different references (e.g., the applicant's IDS or the examiner's citations), the analysis must be reconciled with that list.
2. Claim text is not fully verified. My element mapping of claim 1 relies on a partial reproduction of the granted claim (Justia) plus the specification's Summary. The exact antecedent structure of claim 1 — in particular whether the preamble's "receiving radar data" step is recited in the claim body, and whether the closing step is "calibration" or "extrinsic calibration" — is unconfirmed against the printed patent.
3. Priority/date sensitivity. All § 102(a)(2) characterizations (especially PA‑1 and the contingent PA‑8) assume the RU 2018‑12‑29 priority is imperfect. If the priority is not entitled, the effective date is 2019‑09‑10 and the art set broadens; if the priority is not entitled and the claim scope requires support not present in the Russian application, additional intervening references may qualify.
4. Reference details not independently re-verified. PA‑6's exact issue date, PA‑1's publication date, and EP 3 239 737's US family mappings were taken from the search results shown, not from official register records. Verify each reference's § 102 status and family before relying on this analysis in a filing.
5. This is a technical/analytical assessment, not a legal opinion. An invalidity contention requires claim-construction positions, verified certified copies of each reference, and confirmation of the effective filing date.

Recommended verification steps

1. Obtain the certified printed claim set for US 11,402,467 from USPTO Patent Center / Patent Public Search, and re-run the element mapping against claim 1 verbatim.
2. Confirm the RU priority entitlement (compare RU 2018147498 disclosure against claim 1) to lock the effective date and the § 102(a)(2) window.
3. Pull the IDS and examiner's cited references from the file wrapper — a "Prior Art section" in the intended sense may already contain art that is cumulative of, or narrower than, PA‑1/PA‑2/PA‑3.
4. For PA‑1, verify the effective filing date (2018‑05‑30) and inventorship (Song et al./Wu & Lakhal‑Ayat, different from Sumin) to confirm § 102(a)(2) eligibility.
5. If a formal challenge is contemplated, obtain an English translation of the CN 110554376 A family member and the full text of PA‑2's US counterpart (US 2020/0326411 A1 / US 10,948,569 B2) to align paragraph citations.