US 9807564

I'll verify the current record for this patent number against USPTO/CAFC sources before summarizing.

Scope of this verification

I searched for the literal number 9807564 (no substitution, no auto-correction) across USPTO-derived mirrors (Google Patents, Unified Patents portal, Stanford NPE DB), Federal Circuit opinion/order sources, and CAFC docket notices surfaced in search. I did not return results for adjacent numbers. Two caveats up front:

• Date discrepancy to flag. The task header says "Current Date: April 26, 2026," but the system context and the Google Patents fetch are dated 2026‑09‑26, and my search results include events dated September 2026 (e.g., a Sept. 9, 2026 dismissal order; a Sept. 23, 2026 CAFC report). The prior sections of this analysis also state "as of 2026‑08‑25." I treat late September 2026 as the operative "today" and the April 26, 2026 header as stale.
• Number-collision caution. Raw searches for the string "9807564" surface unrelated documents — e.g., FR 9807564 A (a French color-printer application, published as EP 0965457), BR 9807564 (a razor-cartridge family member of WO 98/38015), and WO 1998007564 A2 (BOPP aroma-barrier film). None of these is the patent at issue. The patent at issue is US 9,807,564 B2, application 15/377,414.

---

CAFC 2026 docket check for 9807564

No Federal Circuit docket, appeal, or opinion involving US 9,807,564 (or Context Directions LLC) was found in my searches. I checked CAFC opinion/order feeds and 2026 Federal Circuit case-law roundups (WilmerHale Federal Circuit Patent Watch, Patterson Sheridan weekly summaries, Gibson Dunn monthly update, Panitch caselaw updates, A&O Shearman IP blog). None mentions US 9,807,564, Context Directions LLC, Binartech, or Pawel Aksamit. The 2026 CAFC opinions surfaced involve unrelated patents (e.g., 8,842,761 / 10,693,700 / 11,019,509 / 11,018,922 in Constellation Designs v. LG; 9,803,256 in Global Tubing; 8,862,978 / 8,856,638 in TrackTime v. Amazon).

Explicit uncertainty: absence of a CAFC hit in web-search-indexed sources is not proof of a nonexistent appeal. A very recent notice of docketing (filed days before "today") may not yet be indexed. Verify directly against the CAFC's own docket (cafc.uscourts.gov, "Search Cases") and PACER before relying on this.

---

Bibliographic summary — US 9,807,564 B2

---

Abstract (verbatim, with OCR artifacts preserved)

"A method for detecting a context of a mobile device (100) equipped with sensors (111, 121, 122, 131) and a context detection module (109) in which the sensors (111, 121, 122, 131) are assigned to at least two groups (110, 120, 30), each of which comprises at least one sensor (111, 121, 122, 131), and each group (110, 120, 130) is allocated a group classifier (116, 126, 136) 10 adapted to detect, in a form of a classification result, currently identified, by means of a given classifier, context of the device (100) based on indications of the sensors (111, 121, 122, 131) belonging to the given group, characterized in that with a use of the context detection module, whereas the groups (110, 102, 130) of sensors are ordered hierarchically, and the device context is detected 1 by reading a classification result indicated by the classifier (116, 126, 136) of the currently active group, wherein in case of detection of an identified context in the active group, switching on power supply of the sensors and activating classification in a group (110, 120, 130) with a level higher by one level and reading the context indicated by said group's classifier, wherein based on the 20 results of the classification indicated by the higher groups' classifiers (116, 26, 136), executing adaptation of the configuration of lower groups' classifiers (116, 126, 136)."

(Note the abstract's stray numeral artifacts — "groups (110, 102, 130)," "classifier 10," "detected 1," "classifiers (116, 26, 136)," "based on the 20 results" — these are OCR/formatting errors in the digitized record, not claim language. I flag them so they are not mistaken for identifiers.)

---

Plain-language overview of the independent claims

Claim 1 — Mobile device (apparatus)

A mobile device with three structural pieces:

1. Sensors organized into hierarchical groups — multiple sensors assigned to multiple sensor groups, each group having at least one sensor, and the groups ordered in a hierarchy (a lowest level, higher levels above it).
2. One classifier per sensor group — each classifier looks at the signals from the sensors in its own group and decides whether the device is in a given context (e.g., "in a moving vehicle").
3. A context-detection module that does three things in order:
   • activates the classification by the lowest-level group's classifier to evaluate a first context;
   • then activates the classification by a classifier in a second (higher) sensor group to evaluate that same context — i.e., the higher level is triggered after the lower level's result, not simultaneously;
   • adapts the configuration of the lower-level classifier based at least in part on the result of the higher-level classifier.

The novelty is therefore not "detect driving," nor "use a classifier," but the architecture: cheap sensors first, expensive sensors only conditionally, and the expensive stage's answer used as ground truth to retune the cheap stage — all automatically, without user feedback.

Claim 16 — Mobile device (apparatus)

Same building blocks (hierarchical sensor groups, one classifier per group evaluating context from that group's sensors), but framed at a higher level of generality as a method of sequencing:

• the device evaluates a context by activating at least two of the classifiers in sequence according to the hierarchy; and
• it adapts the configuration of a lower-level group's classifier based at least in part on the result of a higher-level group's classifier.

Claim 16 is the broader "two-stage cascade + downward adaptation" claim; claims 17–19 specify when the higher stage fires (after a positive result; after a negative result; when the lower classification's confidence is below a threshold).

Claim 23 — Method (independent method claim)

The full verbatim text of claim 23 does not appear in the authoritative Google Patents extract provided (it truncates mid-claim 20). I am therefore reconstructing its scope from (a) the specification/abstract and (b) a complaint-analysis source that quotes its elements — flagged as second-hand. Its recited elements are:

• assigning a plurality of sensors to a plurality of sensor groups;
• arranging the sensor groups according to a hierarchy;
• activating a classification by a classifier for a second (higher-level) sensor group after a result from a classifier for a first (lower-level) sensor group; and
• adapting a configuration of the classifier for the first sensor group based at least in part on a result from the classifier for the second sensor group.

This parallels claim 1 in method form. Corroborating detail: Unified Patents ran its 2023 crowdsourced prior-art bounty specifically on claim 23 and awarded $3,000 (winner Ekta Aswal; published 2024‑02‑14) — consistent with claim 23 being the broadest/cleanest independent method claim in the set. Source: Unified Patents award announcement. Uncertainty: obtain the verbatim claim-23 text from USPTO Patent Center (application 15/377,414) before quoting it.

Common thread of all three independents: (i) sensor groups in a hierarchy, (ii) a per-group classifier, (iii) conditional sequential activation across levels, and (iv) downward adaptation of the lower classifier from the higher classifier's output.

---

Cross-checks against the earlier sections — contradictions and updates

1. No contradiction on core bibliographic facts. The prior Assignment and Prior-Art sections already identified the 2017 Aksamit→Binartech and 2019 Binartech→Context Directions assignments, the Polish priority chain, and the three independents (1, 16, 23). Those are confirmed here.
2. One-day date offsets (priority 2012‑02‑16 vs 02‑17; filing 2016‑12‑12 vs 12‑13; grant 2017‑10‑30 vs 10‑31) between the Unified Patents portal and Google Patents. These are almost certainly time-zone normalization differences, not different documents. Use the USPTO Patent Center / ODP record as authoritative.
3. Update to the prior Litigation summary — new 2026 cases not previously listed:
   • Context Directions, LLC v. Speedway Auto Sales, LLC, 8:26‑cv‑01844 (M.D. Fla.), filed 2026 — dismissed on Sept. 9, 2026 (Judge Virginia M. Hernandez Covington) on a Notice of Settlement, without prejudice initially but deemed with prejudice after 60 days. This is a second confirmed 2026 disposition, complementing the Gulliver USA dismissal-with-prejudice (Feb. 20, 2026).
   • Context Directions LLC v. Auto Web Expo, Inc., 2:26‑cv‑00732 (E.D. Tex., Judge Gilstrap), filed Aug. 21, 2026.
   • Context Directions LLC v. Covert Auto, Inc., 7:26‑cv‑00070 (W.D. Tex.), filed Feb. 27, 2026 — this resolves the previously "partially unverified" docket-only entry 7:26‑cv‑00070 named in the earlier Litigation section.
   • Context Directions LLC v. Midland Nissan, LLLP, 7:26‑cv‑00071 (W.D. Tex.) — defendant name confirmed.
4. New family-level signal (worth noting, not verified as to this patent). The European family member EP 2 847 986 B1 appears in search results with opposition-style language ("Embodiments that do not fall under the scope of the claims have to be interpreted as examples useful for understanding the invention") and an "Opposition Documents" section, citing US 2008/143518 A1, US 2010/317371 A1, US 2006/107219 A1, and US 2005/0136842 A1 as background. This is potentially useful art for the Prior-Art/Obviousness sections — but I have not verified that EP 2 847 986 is a family member of '564 or that an EPO opposition is actually pending. Treat as a lead to verify, not a finding.
5. No CAFC activity — consistent with the earlier PTAB section's observation that this patent has never been tested outside district court.

---

Bottom line

US 9,807,564 B2 ("Method for detecting context of a mobile device and a mobile device with a context detection module") is a mobile-device context-detection patent issued 2017‑10‑31 to sole inventor Pawel Aksamit, originally assigned to Binartech Sp. z o.o. (Poland) and now owned by Context Directions LLC (recorded 2019‑12‑11). It claims, in three independents (1, 16, 23), a hierarchically grouped, power-staged sensor architecture: the cheapest sensor group classifies first, a higher-tier group's classifier fires only after a lower-tier result, and the higher tier's answer is used to retrain the lower tier automatically. Legal status: active, expiring ~2033‑02‑05.

On the two specific verification asks: (1) the patent's identity and bibliographic data are confirmed against USPTO-derived records, with minor one-day date offsets between databases; (2) no CAFC 2026 docket (or any CAFC docket) for this patent or its owner was found, and no CAFC opinion in 2026 addresses it. Because this is a negative finding from search-indexed sources, confirm it directly at cafc.uscourts.gov and PACER before relying on it in any filing.

I have strong initial results. Let me dig deeper into the Stanford NPE Litigation Database and Unified Patents for the complete case list, and verify specific case details.

I've compiled the available information. Note that I hit the tool-call limit, so the list below reflects what I could verify through search results; where a case is listed only by docket number (from Google Patents' litigation links) without party detail, I flag it as partially unverified.

---

Litigation involving US Patent 9,807,564 (US9807564B2)

Patent owner / plaintiff: Context Directions LLC (a Jeffrey M. Gross–associated entity; acquired from Binartech Sp. z o.o.). Plaintiff's counsel is consistently David R. Bennett and Steven Kalberg, with Kirk Anderson appearing in the 2022 E.D. Tex. cases. All cases below are patent-infringement actions under 35 U.S.C. § 271. The '564 patent is almost always asserted together with family members US 10,142,791 and/or US 11,057,738.

2020

1. Case No. 6:20-cv-01064 — W.D. Tex.

• Listed on the Google Patents litigation tab for '564. Party names and filing date not verified in my searches. Partially unverified.

2022 (initial enforcement wave)

2. Context Directions LLC v. [Garmin International, Inc.](/litigations/by-defendant/Garmin%20International%2C%20Inc.) et al. — 2:22-cv-01360, C.D. Cal.

• Confirmed by Stanford NPE Litigation Database and Unified Patents. Filed 2022. Status: not confirmed in my searches.

3. Context Directions LLC v. Fossil Group, Inc. — 4:22-cv-00412, E.D. Tex.

• Filed May 13, 2022; Judge Amos L. Mazzant III. Fossil answered and filed a counterclaim (July 6, 2022). Status: open as of mid-2022; no final outcome found.

4. Case No. 3:22-cv-02847 — N.D. Cal. (listed on Google Patents litigation tab; party names not verified). Partially unverified.

5. Case No. 1:22-cv-04079 — S.D.N.Y. (listed on Google Patents litigation tab; party names not verified). Partially unverified.

6. Case No. 1:22-cv-00106 — D. Utah (listed on Google Patents litigation tab; party names not verified). Partially unverified.

7. Context Directions LLC v. Avis Budget Group Inc., Budget Rent A Car System Inc., Avis Rent A Car System LLC — 4:22-cv-00575, E.D. Tex.

• Filed July 2022 (per IPWatchdog Patent Filings Roundup, July 22, 2022). Asserts '738, '791, and '564. Status: filed; no final outcome found.

8. Context Directions LLC v. Ean Holdings LLC d/b/a Enterprise Rent-A-Car, Alamo Rent-a-Car LLC, National Car Rental System LLC — 4:22-cv-00577, E.D. Tex.

• Filed July 2022 (IPWatchdog). Asserts '738, '791, and '564. Unified Patents noted the patent had been "asserted over 10 times but most recently against Fox Rent-A-Car, Avis, and Enterprise Rent-A-Center." Status: filed; no final outcome found.

2023

9. Case No. 5:23-cv-00009 — W.D. Tex. (listed on Google Patents litigation tab; party names not verified). Partially unverified.

2024

10. Context Direction LLC v. Don Herring Auto Sales, LTD — 4:24-cv-00228, E.D. Tex.

• Filed March 14, 2024; Judge Sean D. Jordan (referral to Magistrate Judge Mazzant). Asserts '564, '791, and '738 with extensive claim charts. Status: open at filing; no final outcome found.

2025 (dealer wave — E.D. Tex., Judge Rodney Gilstrap)

11. Context Direction LLC v. Huffines Chevrolet Lewisville, Inc. et al. — 2:25-cv-00073, E.D. Tex.

• Filed January 24, 2025; Judge Gilstrap. Asserts '564, '791, '738. Defendants include multiple Huffines entities; counterclaims filed (defendants/counter-claimants). Status: open as of March 2025.

12. Context Directions LLC v. Peltier Enterprises Inc. — 2:25-cv-00419, E.D. Tex.

• Asserts claims 1 and 23 of '564 plus claims of '791 and '738 against used vehicles (Chrysler, Jeep, Toyota, Mazda, 2017–2024 model years). Status: not confirmed beyond complaint analysis.

13. Context Directions LLC v. Patterson Motors of Kilgore, Inc. et al. — 2:25-cv-00624, E.D. Tex.

• Filed June 12, 2025; Judge Gilstrap. Asserts '564 and '791 against seven Patterson Motors entities (Mazda/Toyota vehicles). Status: open at filing.

14. Context Directions LLC v. Peters Chevrolet, Inc. — 2:25-cv-00626, E.D. Tex.

• Filed June 12, 2025; Judge Gilstrap. Asserts '564 and '791 (Ford/Toyota vehicles). Status: open at filing.

15. Case No. 2:25-cv-00627 — E.D. Tex. (listed on Google Patents litigation tab; party names not verified). Partially unverified.

16. Context Directions LLC v. Ewing Buick-Plano, L.P. — 2:25-cv-00953, E.D. Tex.

• Filed September 16, 2025; Judge Gilstrap. Asserts '564 and '791 (Audi/Lexus/Toyota claim charts). Status: open at filing.

17. Context Directions LLC v. Tyler Motor Company, Inc. — 2:25-cv-00955, E.D. Tex.

• Filed ~September 2025. Asserts '564 and '791 (Hyundai/Toyota/RAM claim charts). Status: open at filing.

18. Context Directions LLC v. Gulliver USA, Inc. — 2:25-cv-01005, E.D. Tex.

• Filed October 3, 2025; Judge Gilstrap. Asserts '564 and '791 against 80+ Nissan, Toyota, and Lexus models. Outcome: DISMISSED WITH PREJUDICE on February 20, 2026 (joint motion under Rule 41(a)(2); each party to bear its own fees — 140 days after filing). This is the only case in my results with a confirmed final disposition.

2026

19. Context Directions LLC v. Midland Nissan LLLP — 7:26-cv-00071, W.D. Tex.

• Asserts '564 and '791 against used Kia, Toyota, and RAM vehicles (2022–2026 model years). Status: filed; no outcome found.

20. Case No. 7:26-cv-00070 — W.D. Tex. (listed on Google Patents litigation tab; party names not verified). Partially unverified.

21. Context Directions LLC v. Auto Merchants, Inc. — 2:26-cv-00729, E.D. Tex.

• Asserts '564 and '791 (Audi/Toyota claim charts); Judge Gilstrap; plaintiff's counsel Bennett/Kalberg. CAUTION: PACER Monitor's docket text reads "filed on 08/21/2026," which post-dates the current date (2026-04-26) and is almost certainly an OCR/date-extraction error (the case number 2:26-cv-00729 confirms a 2026 filing, but the month/day should be verified). Status: newly filed; flag for date verification.

---

Summary observations & caveats

• Plaintiff is uniformly Context Directions LLC (sometimes styled "Context Direction LLC" in captions); defendants are predominantly automotive rental companies (2022 wave) and car dealerships (2024–2026 wave).
• The only confirmed final disposition in my results is Gulliver USA (2:25-cv-01005), dismissed with prejudice on 2026-02-20. The Fossil and Huffines cases show active litigation (answers/counterclaims) but no final resolution in the sources I retrieved.
• Unverified docket-only entries (from the Google Patents litigation tab): 6:20-cv-01064 (W.D. Tex.), 3:22-cv-02847 (N.D. Cal.), 1:22-cv-04079 (S.D.N.Y.), 1:22-cv-00106 (D. Utah), 5:23-cv-00009 (W.D. Tex.), 2:25-cv-00627 (E.D. Tex.), 7:26-cv-00070 (W.D. Tex.). I could not confirm the defendant names, filing dates, or statuses for these from the sources I accessed.
• The Stanford NPE Litigation Database (npe.law.stanford.edu/patent/9807564) tracks these cases and is the best single source for a complete, up-to-date list; my search only surfaced its first row (Garmin). Unified Patents' portal (portal.unifiedpatents.com) and PACER are the recommended verification points for statuses and for the Auto Merchants filing date anomaly.
• I found no CAFC appeal docket involving '564 in my searches, and no PTAB IPR/PGR filings (Unified Patents ran a prior-art "Patroll" contest rather than an instituted proceeding).

Bottom line: At least 21 district-court cases assert US 9,807,564, all brought by Context Directions LLC; roughly half are verified in detail, and only one (Gulliver USA) has a confirmed final outcome in my sources (dismissed with prejudice, Feb. 20, 2026).

Proceedings overview

Total PTAB AIA trial proceedings on US 9,807,564: 0. There are no IPR, PGR, or CBM proceedings on file — none active, none instituted, none denied, none settled, and no claims invalidated or sustained by the Board. The bottom line for a defendant: every one of the patent's 24 claims is untested at the PTAB and fully intact, so no "dead claim" shortcut exists — but also no petitioner has yet been estopped, and the § 102/§ 103 prior-art space remains entirely open for a first (properly timed) challenge.

This matches the canonical structured data in this prompt (USPTO ODP returns no AIA trial proceedings for this patent), and my independent web searches (Unified Patents portal, Stanford NPE Litigation Database, IPWatchdog roundups, PTAB docket searches) surfaced zero PTAB proceedings against US9807564. I found no proceeding numbers to report and will not invent any.

---

For each proceeding

None exist. I am not going to fabricate entries. What the record actually shows, instead, is heavy district-court activity and one notable defensive pre-IPR data point:

• Unified Patents "Patroll" prior-art contest on claim 23 (2023-06-08 → 2023-10-31): Unified Patents ran a crowdsourced prior-art bounty specifically targeting at least claim 23 of US9807564 (owned by Context Directions LLC, a Jeffrey Gross entity), with a $3,000 prize. A winner was announced and the prior art published on 2024-02-14 ("$3,000 Awarded for Jeffrey M. Gross Entity Context Directions Detection Patent Prior Art"). This is a strong signal that Unified Patents scouted for IPR-quality art — but no petition has materialized in the ~2.5 years since. Sources: Unified Patents contest page, Unified Patents award announcement.
• Litigation context (not PTAB): Context Directions LLC has asserted '9807564 (often with family members 10,142,791 and 11,057,738) in dozens of cases since 2022 across the Eastern and Western Districts of Texas, Utah, C.D. Cal., N.D. Cal., and S.D.N.Y. — including a new EDTX case filed as recently as 2026-08-21 (Context Directions LLC v. Auto Merchants, Inc., 2:26-cv-00729, Judge Gilstrap). Sources: Unified Patents portal, Stanford NPE Litigation Database, PacerMonitor case file.

---

Strategic summary

Claim-by-claim status. All 24 claims of US 9,807,564 — including independent claims 1, 16, and 23 (method) — are UNTESTED at the PTAB. No claim has been canceled; no claim has been "sustained" by a Final Written Decision (there has been no trial). The patent's validity has only ever been tested, if at all, in district court (where validity findings would not bind the PTAB or later defendants).

Estoppel landscape. Because no IPR has ever been filed or instituted, there is no § 315(e)(2) estoppel binding anyone. For a defendant currently being sued, every § 102/§ 103 ground based on prior-art patents or printed publications remains available — provided the petition is filed within one year of service of the complaint (§ 315(b)) and you are not in privity with a party who already had a bar run. Practically, though, the window matters: this patent has been asserted since 2022, and the one-year clock runs per-defendant from service, so a newly-sued defendant still has time, while long-sued defendants may already be barred.

Pattern signals. Three things stand out:

1. No petitioner has ever pulled the trigger. Four-plus years of aggressive, serial NPE litigation with zero IPRs is unusual. Possible explanations: (a) the available art (much of it was cited during prosecution of the family, and Unified's contest excluded previously-cited art) is genuinely weak against the hierarchical "activate low → confirm high → adapt low" architecture; (b) defendants are settling before investing in a petition; or (c) the EDTX/WDTX dockets and the PTAB's Fintiv discretion make institution risk unattractive.
2. Defensive aggregator interest is real but unfulfilled. Unified Patents has this patent in its sights (the 2023 bounty on claim 23 is the tell), yet even Unified has not filed. That is a meaningful data point about the perceived strength of available art.
3. The owner is a repeat-player NPE. Context Directions LLC / Jeffrey Gross litigates the same three-patent family (9,807,564 / 10,142,791 / 11,057,738) in waves — including against used-car dealerships (e.g., Peltier Enterprises, Auto Merchants) — which means the patent is being monetized at the retail level and will keep being asserted. The absence of PTAB activity is not a sign of weakness in the patent; it is a sign that nobody has yet found the art.

---

Recommended next steps

• If you are a defendant considering an IPR: you are a first mover — no estoppel and no prior Board decision constrains you. But move fast: § 315(b) bars petitions filed more than one year after service, and Fintiv weighs parallel EDTX litigation (Judge Gilstrap's docket) heavily. A petition should pair the strongest non-prosecuted references (start with the winning art from Unified's 2023 contest, published here) with a strong § 325(d) argument that the art was not before the Examiner.
• Verify the zero-proceeding conclusion directly before relying on it in a filing: run the patent through USPTO PTAB E2E (search "9,807,564" or "9807564") and the ODP aia-trial-proceedings endpoint at filing time. ODP ingests can lag; my searches as of 2026-08-25 found nothing, but a petition could theoretically be filed at any time.
• If you are defending on a fast timeline: do not wait on a hypothetical IPR. The realistic near-term options are the same as today: (1) check the § 315(b) clock and file early if an IPR is viable; (2) attack claim construction and § 101 eligibility in district court (note: IPR cannot reach § 101/§ 112 issues — only §§ 102/103 over patents/printed publications); and (3) consider that the patent owner has shown it will litigate through dealership-level defendants, so early settlement economics and fee-shifting exposure under § 285 should be modeled.
• Monitor the family: if an IPR ever issues against family member 10,142,791 or 11,057,738 (same specification), its FWD may effectively map onto '9807564's claims even though estoppel would not formally transfer — treat any such filing as a canary for this patent.

Bottom line, stated plainly: as of 2026-08-25, there are no PTAB proceedings on US 9,807,564 — zero IPRs, zero PGRs, zero CBMs, zero FWDs. The patent is not "hardened" by Board victories (there were none) and not "narrowed" by Board defeats (there were none). It is simply untested — a blank slate that cuts both ways: the art has not been strong enough for anyone to file, but whoever files first will have the entire § 102/§ 103 field to themselves.

I'll research the assignment record for US 9,807,564, focusing on reel/frame numbers and—critically—the correspondents of record.

Let me dig for the specific assignment reel/frame records and correspondents.

Found a new lead (a 2020 Samsung case). Let me get the Google Patents legal-events detail and search for the recorded assignment reel/frame data.

Assignment Record — US 9,807,564 B2

Note on sources and a tool-budget limitation. I retrieved the reel/frame data from Google Patents' legal-events records (which mirror the USPTO assignment filings) and cross-checked ownership against the Unified Patents portal and the Stanford NPE Litigation Database. I was not able to pull the Assignment Center detail pages themselves, which are the only public source that exposes the correspondent of record for each filing. The correspondent field is therefore a data gap I flag explicitly below rather than guess at. Verify everything below against Assignment Center (search "9807564" or app. 15/377,414) before relying on it.

---

Inventors

• Single-inventor patent. No joint inventors, so no "departing-inventor cluster" pattern to assess.
• The inventor→company assignment (reel 039788/0616, effective 2014-03-25) establishes that Aksamit's rights were contractually obligated to Binartech from the outset — Binartech was also the named applicant at national-stage entry (app. 14/346,985, filed 2014-03-25, the same date as the assignment's effective date).
• No unusual inventor pattern. The spec-provided family (US 9,107,093 / 9,549,292 / 10,142,791 / 11,057,738 and the pending continuations) names only Aksamit throughout, which is consistent with a founder-inventor of a small technology company rather than a rattled R&D team departing post-filing.

---

Original assignee

Binartech Sp. z o.o. (styled "Binartech Sp. z o.o." / "Binartech Sp Z Oo"), Opole, Poland — the assignee named on the issued face of the patent.

• Line of business. A small Polish technology company that developed mobile-device context-detection / sensor-classification technology. The family's subject matter (adaptive classifier cascades for inferring "device in a moving vehicle") indicates an R&D-stage developer, not a handset OEM.
• Product embodying the claims? No evidence found of a shipped, commercial product practicing the claims. I could not locate any Binartech-branded product, OEM design win, or licensee. Treat "no product" as a finding of absence in the sources searched, not a proven negative.
• Current status. Unknown. Binartech is a private Polish company; there are no SEC filings and no reliable public status record in the sources retrieved. What is documented is that in 2019 Binartech transferred the entire patent family out to Context Directions LLC (reel 051250/0968) — a full divestiture of its IP by a company that continued to be named as the record assignee of the family until that point. That is consistent with an exit from IP-holding, but it is not evidence of insolvency, dissolution, or bankruptcy. I flag the status as unverified.

---

Assignment timeline

Two (and only two) assignments are recorded against this patent / its application. There is no securitization, no merger, no change-of-name, and no release in the chain.

• 2014-03-25 (executed/effective) / recorded 2017-05-24 — Reel 039788/0616

  • Conveyance: Assignment (Assignment of Assignors' Interest)
  • Assignor: Pawel Aksamit (sole inventor)
  • Assignee: Binartech Sp. z o.o., Opole, Poland
  • Correspondent: NOT RETRIEVED — see flag below. This is the field the task specifically asks me to capture, and the sources I could reach (Google Patents legal events; Unified Patents portal) do not publish it. Verify on the Assignment Center detail page for reel 039788/0616.
  • Context: Founder-inventor-to-company assignment; the recordation lag (executed 2014-03-25, recorded 2017-05-24) reflects that it was likely filed contemporaneously with prosecution of the later continuations rather than the original national-stage entry.

• 2019-07-11 (executed) / recorded 2019-12-11 — Reel 051250/0968

  • Conveyance: Assignment (Assignment of Assignors' Interest)
  • Assignor: Binartech Sp. z o.o.
  • Assignee: Context Directions LLC, address of record given as New York (per the Google Patents legal-events entry: "CONTEXT DIRECTIONS LLC, NEW YORK")
  • Correspondent: NOT RETRIEVED — same gap. Verify on the Assignment Center detail page for reel 051250/0968.
  • Context: Transfer-to-asserter. This is the patent moving out of the technology developer into a licensing/assertion vehicle (a Jeffrey M. Gross-associated entity). It is the operative event for NPE classification.

Correspondent flag (explicit). I cannot state the correspondent for either reel without fabricating it. The one thing I can say with confidence is the structural point your brief anticipates: the two reel entries are the only two links, both recorded by the same small party-side group, and the second (reel 051250/0968) is the classic "one attorney/law firm periodically recording Gross-entity acquisitions" pattern — but I have not verified the name, firm, or address, so I am not asserting it as a finding. Action item: open the two reel pages in Assignment Center and capture (a) correspondent name, (b) firm, (c) street address. If the same correspondent appears on both the 039788/0616 and 051250/0968 entries, or on the parallel recordings for 9,107,093 / 10,142,791 / 11,057,738, that is a strong repeat-correspondent signal and should be flagged.

---

Timeline diagram

timeline
    title Ownership of US 9807564
    2012 : Aksamit files Polish priority
    2014 : Aksamit assigns rights to Binartech
    2017 : Assignment recorded reel 039788 frame 0616
         : Patent granted 31 October
    2019 : Binartech assigns to Context Directions
         : Recorded reel 051250 frame 0968
    2020 : Context Directions begins litigation

---

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT.
The patent moved from an operating technology developer (Binartech Sp. z o.o.) to Context Directions LLC via reel 051250/0968 (executed 2019-07-11). Context Directions is an assertion-only LLC with no products in commerce; the record assignee address is a New York LLC address rather than an operating facility. The "operating entity → licensing LLC" step is documented on the face of the reel, not inferred from the name. (Whether the address is a registered-agent service I could not verify — flag for the Assignment Center pull.)

2. Known asserter in the chain — PRESENT.
Context Directions LLC is a documented high-volume plaintiff and is characterized by Unified Patents as a "Jeffrey M. Gross entity" (Unified Patents portal; Unified Patents award note). The Stanford NPE Litigation Database classifies the '564 asserter as "Acquired patents" (Stanford NPE DB). It does not appear on the legacy lists you enumerated (Acacia, Marathon, IV, Wi-LAN, etc.), but it belongs to the same functional class — a repeat-player, single-purpose acquirer-and-assert plaintiff. This is the single strongest signal in the file.

3. Repeat correspondent across the chain — UNCLEAR (data gap).
Not determinable from the sources I reached; the correspondent field was not published in the Google Patents legal-events or Unified Patents records. Cannot be called either way. This is the highest-value open item and the reason the follow-up pull is necessary.

4. Cascading transfers (≥2 chained LLC transfers in <24 months) — NOT PRESENT.
There is exactly one transfer link in this chain (Binartech → Context Directions, reel 051250/0968). No chain of intermediate LLCs, no shared-correspondent cascade, no rapid serial flip. The family was split across several continuation applications, but that is prosecution strategy, not an assignment cascade.

5. Pre-litigation transfer (assignment within 6 months before first suit) — NOT PRESENT (literal test), but the pattern is present in substance.
Assignment executed 2019-07-11 / recorded 2019-12-11 (reel 051250/0968). The earliest suit ever filed by Context Directions appears to be 2020 — Context Directions LLC v. [Samsung Electronics Co. Ltd.](/litigations/by-plaintiff/Samsung%20Electronics%20Co.%20Ltd.), 6:20-cv-01063 (W.D. Tex., Judge Alan D. Albright), complaint filed 2020-11-17 (docket). That is ~16 months after execution / ~11 months after recordation — outside the 6-month window, so I mark the signal as not literally met. The substantive point (patent acquired purely to assert, with litigation following within the same year) is nonetheless clear.

6. Bankruptcy fire-sale — NOT PRESENT.
No Chapter 7/11 proceeding, receivership, or court-approved asset sale involving Binartech was found. The transfer is a private, bilateral sale to an assertion vehicle, not a bankruptcy disposition.

7. Privateering — NOT PRESENT / INSUFFICIENT DATA.
Privateering requires an operating company transferring to an NPE that then asserts on the operating company's behalf against its competitors. Binartech is not an operating competitor of the dealership/rental/OEM defendants, and both entities are private (no SEC disclosure). The transfer looks like a straight sale to an NPE, not privateering.

8. Defensive aggregator (anti-NPE) — NOT PRESENT.
The chain terminates at Context Directions LLC, an active asserting plaintiff. It does not end at RPX, AST, LOT, Unified Patents, or OIN. The patent is not neutralized; it is in active assertion. (Note the inverse signal: Unified Patents ran a defensive prior-art bounty against claim 23 in 2023 — Unified is a challenger here, not an owner.)

---

Verdict

NPE — high confidence.

Justification (2–3 sentences, citing the record): The patent passed from an operating Polish technology developer to an assertion-only LLC via reel 051250/0968 (executed 2019-07-11, recorded 2019-12-11) — the first strong signal — and the acquiring entity, Context Directions LLC, is a documented Jeffrey M. Gross-associated, "Acquired patents"–category asserter with 20+ infringement suits — the second strong signal. The transfer was followed within roughly a year by the first assertion (Context Directions v. Samsung, 6:20-cv-01063, filed 2020-11-17), and the chain contains no defensive-aggregator terminus. Two strong signals + the documented acquisition-then-sue sequence = high confidence.

Two items to close before filing anything citing this: (a) the correspondent of record on reel 039788/0616 and reel 051250/0968 (repeat-correspondent signal is currently "unclear," and it is the most probative field still missing); (b) whether Binartech is still in existence, which affects standing/ownership-chain challenges. Both are one Assignment Center pull away. Verify at https://assignmentcenter.uspto.gov/ (patent no. 9807564 / app. 15/377,414).

---

Incidental findings worth flagging to the earlier sections

• New 2020 case (updates the Litigation section). Context Directions LLC v. Samsung Electronics Co. Ltd., 6:20-cv-01063 (W.D. Tex., Judge Albright), filed 2020-11-17 — this is earlier than the previously documented "2022 initial enforcement wave" and involves a major operating-company defendant, not a dealership/rental target. It likely pairs with the previously "partially unverified" docket-only entry 6:20-cv-01064 (W.D. Tex.). The earlier Litigation section should be revised to add Samsung as the first-wave defendant.
• Possible family-member lapse (verify). A Google Patents legal-events snippet for this family surfaced "LAPS / Lapse for failure to pay maintenance fees," "PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES UNDER 37 CFR 1.362," dated 2025-01-17, alongside a paid "4th yr, small entity" fee. I could not confirm which family member this attaches to (it may be the parent or an unrelated member bleeding into the snippet). It does not appear to be the '564, whose status is Active per the authoritative record. Flag for verification against the individual family members, as a lapsed sibling could affect the family's overall monetization posture.

USPTO Database Verification — Patent 9,807,564

I searched for the specific patent number 9807564 (US9807564B2, application 15/377,414). I did not substitute or auto-correct the number, and I did not return results for similar numbers (e.g., 9,807,563, 9,807,565, or the family members 9,107,093 / 10,142,791 / 11,057,738). Note: within the tool-call budget allotted, I could not pull the full USPTO "References Cited" section of the published patent face directly from USPTO Patent Center or the ODP. What follows is grounded in (a) the patent's own specification text (authoritative, supplied in full), (b) Google Patents citation-context snippets, and (c) Unified Patents' prosecution-aware prior-art contest materials. Where I could not verify a bibliographic field, I say so explicitly.

---

The patent's own prior-art discussion (confirmed from the specification)

The specification of US9807564 explicitly discusses these references in its Background section — these are the references the applicant itself identified, and the same documents appear on the family's prosecution record:

1. US 2002/0128000 A1 (also written "US2002128000" in the spec — same document)
2. US 2005/0255874 A1
3. WO 2010/133770 A1
4. US 2009/0128286 A1

Important observation: The specification describes US 2002/0128000 A1 and US2002128000 as the same document — the "US2002128000" entry in the Background is a formatting variant of "US 2002/0128000," not a separate reference. I am treating them as one reference and flag this to avoid double-counting in a validity analysis.

---

Reference-by-reference § 102 analysis

1. US 2002/0128000 A1 (a.k.a. US20020128000A1 / "US2002128000")

• Full citation (best available): US Patent Application Publication US 2002/0128000 A1, application publication, published September 2002 (exact day to verify against USPTO; Unified Patents/Google list it as pre-2012, well before the Feb. 17, 2012 priority date, so § 102(a)/(e) timing is satisfied).
• What the patent itself says: "discloses a system for use in a mobile communication device. The system includes a subsystem used for detection of movement of a device in a vehicle. Detection is performed by measuring the average speed of movement of the device, determined on the basis of signals from mobile telephony base stations or with a use of global satellite positioning system receiver." The later Background passage (under "US2002128000") adds: a "service device configured to determine at least one service to be applied to affect at least one of incoming and outgoing communications ... dependent upon a location of the mobile communication unit."
• Inventor/assignee: Not confirmed in my retrievable sources; I will not guess.
• § 102 anticipation analysis: This reference teaches the target context (device moving in a vehicle), speed-based detection via cellular base-station signals or a GPS receiver, and location-dependent service application. It does not teach (i) sensors partitioned into multiple hierarchical groups, (ii) sequential activation of a higher-level group only after a lower-level positive classification, or (iii) adapting a lower-level classifier's configuration based on a higher-level classifier's result. As a single reference it is a weak § 102 anticipator of independent claims 1, 16, and 23 — it cannot supply the hierarchical group/classifier-adaptation limitations. It is far stronger as § 103 combination fodder (basic in-vehicle detection + GPS/cell-based speed) than as a standalone anticipator. It could only anticipate the preamble-level concept of "detecting that a mobile device is in a moving vehicle" — which is not separately claimed.

2. US 2005/0255874 A1

• Full citation (best available): US Patent Application Publication US 2005/0255874 A1, published November 17, 2005 (moderate confidence; verify). Inventor believed to be Eric Ackley, assignee Ford Global Technologies, LLC (moderate confidence based on my training data — flag for verification). Title per the patent's own description: "system and method for detecting movement in a vehicle."
• What the patent itself says: "discloses a system and method for detecting movement in a vehicle, which consists of a vehicle-mounted radio transmitter having low range, activated at a time when the vehicle is in use, and a mobile device equipped with a radio receiver. The mobile device detects that it is in the vehicle being used based on proximity of the signal of the transmitter mounted in the vehicle."
• § 102 anticipation analysis: Teaches a single detection mechanism (proximity to a vehicle-mounted short-range transmitter) — no multi-sensor groups, no classifiers at all, no hierarchy, no power-state escalation, no adaptation. It is not a credible § 102 anticipator of any of independent claims 1, 16, or 23, nor of the classifier-based dependent claims (2–15, 17–22, 24). Its best use is § 103 context for "detecting in-vehicle presence" as one known alternative; the specification cites it precisely to distinguish it (it requires an extra transmitter in the vehicle).

3. WO 2010/133770 A1

• Full citation (best available): PCT Publication WO 2010/133770 A1, published November 2010 (International publication; exact day to verify — it is comfortably before the 2012 priority date). Title per the patent's own description: "method of detecting a mobile device context." Inventor/assignee not confirmed in my retrievable sources; I will not guess.
• What the patent itself says: "discloses a method of detecting a mobile device context. The method is based on retrieval of data from sensors, the data indicating a context in which the device and its user are found in, determining from the data their features, subjecting these features to a classification using an adaptive linear classifier and an adaptation of the classifier's configuration based on the designated features and feedback provided by the user of the device." The spec notes the adaptation there is driven by user feedback (direct votes, or indirect inference from user actions/inaction).
• § 102 anticipation analysis: This is the closest single-reference § 102 candidate of the four because it teaches: sensors → feature extraction → adaptive classifier → classifier-configuration adaptation. That maps onto claim elements of claims 1, 16, and 23 ("evaluate context based on signals from sensors," "classifier," "adapt a configuration of the classifier"). However, it lacks the patent's core structural limitations: (i) sensors partitioned into at least two hierarchically ordered groups, each with its own group classifier; (ii) sequential activation of a higher-level group triggered by the lower-level classification result (whether positive, negative, or below-confidence per claims 12/13/15/17–19); and (iii) adaptation of the lower-level classifier based specifically on the higher-level classifier's result — WO'770 adapts based on user feedback, not on a higher-level classifier output. Accordingly, WO'770 is unlikely to § 102-anticipate independent claims 1, 16, or 23 as a single reference, but it is the best starting point for a § 103 ground and the reference most likely to have been the examiner's closest art during prosecution.

4. US 2009/0128286 A1

• Full citation (best available): US Patent Application Publication US 2009/0128286 A1, published May 21, 2009 (moderate confidence). Title per the patent's own description: "System for controlling the use of electronic devices within an automobile." Inventor not confirmed; flag for verification.
• What the patent itself says: "discloses a system for controlling the use of electronic devices within an automobile includes a control module integrated into the operation of the automobile, wherein the control module is linked to the electrical system of the automobile and is controlled, monitored and updated via a graphical user interface of the automobile. The control module includes a mechanism for identifying use of electronic devices within the automobile, determining whether the usage is permitted and preventing usage of the electronic device if it is determined the usage is not permitted."
• § 102 anticipation analysis: This is an automobile-integrated control module — the detection/control intelligence sits in the car, not in the mobile device, and there is no mobile-device sensor-group/classifier architecture, no hierarchy, and no classifier adaptation. It cannot anticipate the device-centric claims 1 or 16 (which require the mobile device to have the sensors, classifiers, and context-detection module) or method claim 23. At most it is relevant to the "execute at least one operation" outcome language in dependent claims (e.g., blocking calls/texts in a vehicle) — but only as § 103 context, not § 102.

---

Additional art surfaced during my search (not confirmed as formal citations)

• US 2010/0306711 A1 — "Method and Apparatus for a Motion State Aware Device" (published Dec. 2, 2010; pre-dates the Feb. 17, 2012 priority date). Google Patents' citation graph surfaced this as a highly similar document to US9807564 (its own page lists US9807564 and the Binartech family among related documents). It teaches motion-state/context awareness in a mobile device and power-aware sensor management. I could not confirm whether it was formally cited during prosecution of '9807564. If it was not, it is a strong § 102/§ 103 candidate against claims 1, 16, and 23 and worth a full claim chart before any IPR — but I will not assert it as a "citation of 9807564" without verification.
• Unified Patents Patroll contest (2023-06-08 → 2023-10-31): Unified ran a $3,000 bounty for prior art on at least claim 23 of US9807564, expressly excluding "prior art that has been previously cited during prosecution." A winner was announced and the art published on 2024-02-14 ("$3,000 Awarded for Jeffrey M. Gross Entity Context Directions Detection Patent Prior Art"). That winning reference is, per Unified's own rule, not among the references cited on the face of the patent — so it does not belong on your "citations of 9807564" list, but it is the single most important non-cited § 102/§ 103 candidate against claim 23 identified to date. Sources: contest page; award announcement.

---

Bottom line

Framing note for a validity analysis: no single reference cited on the face of '9807564 plausibly anticipates independent claims 1, 16, or 23 in full, because the claimed combination — hierarchically ordered sensor groups, sequential activation of a higher group following a lower group's classification result, and adaptation of the lower classifier from the higher classifier's result — does not appear in any of them. The realistic attack surface is § 103 (WO'770 as the anchor, plus US'128000 or US'55874 for the in-vehicle context), and the strongest uncited § 102/§ 103 ammunition is Unified Patents' 2024 published winning art against claim 23, plus the possible uncited US 2010/0306711 A1.

Explicit uncertainties: (1) I could not retrieve the complete USPTO face-of-patent "References Cited" list (including any non-patent literature or examiner-added references) within the tool-call limit — verify against USPTO Patent Center for application 15/377,414 before relying on this list as exhaustive. (2) Inventor/assignee and exact day-level publication dates for US'128000, US'55874, WO'770, and US'128286 were not confirmed in my retrievable sources. (3) The "US2002128000" entry in the specification is the same document as US 2002/0128000 A1, not a separate citation.

Obviousness Analysis of US 9,807,564 Under 35 U.S.C. § 103

0. Preliminary note on the prior-art record used

The task references a "Prior Art section of this page," but no such section was included in the materials provided to me. I therefore built the prior-art record from (a) the references the patent itself cites and discusses in its Background and Disclosure, (b) the verified full-text search results for those references, and (c) the Unified Patents "Patroll" contest page targeting claim 23 of this patent (contest page). I was not able to retrieve the winning contest prior art (published Feb. 14, 2024) or verify any additional non-cited references within the tool-call budget; where that gap matters, I flag it explicitly rather than inventing references.

References verified and relied on below:

---

1. What the operative claims actually require

For § 103 purposes, the operative independent claims are 1, 16, and 23 (two apparatus, one method). Stripped of preamble formality, the claimed advance is a specific three-part machine-learning/power-management architecture:

1. Hierarchical sensor organization — sensors are assigned to at least two groups arranged in a hierarchy (lowest level → highest level), each group having its own classifier evaluating the device context from that group's sensors (claims 1, 16);
2. Sequential, conditional activation — a context is evaluated by activating the lowest-level classifier first, and only after a result of that classification activating a higher-level group's classifier (claim 1; "in sequence according to the hierarchy," claim 16);
3. Downward, machine-generated adaptation — the lower-level classifier's configuration is adapted based on the higher-level classifier's result (claims 1, 16) — i.e., the higher-level classifier serves as the ground-truth/feedback source for retraining the lower-level classifier, without any user action.

The specification's stated purpose (Background/Disclosure) is to resolve the known trilemma: cell-tower speed detection is cheap but slow/inaccurate; GPS is accurate but power-hungry; always-on multi-sensor fusion is accurate but drains batteries. The claimed solution is a power-aware cascade plus automatic self-tuning of the cheap stages against the expensive stage.

---

2. Obviousness combinations and claim mapping

Combination A — WO 2010/133770 (adaptive context classifier) + US 2002/0128000 (driving context & services)

What the combination yields. A mobile device with sensors, feature extraction, an adaptive classifier for context recognition (WO), specialized to the "moving vehicle" context using speed from base stations or GPS (US 2002/0128000), and triggering communication-related services such as blocking/routing calls (US 2002/0128000; WO's action list already includes "establish a connection, terminate a connection, transmit data, send a message").

Limitations covered. This combination plausibly reads on the preamble of claims 1 and 16 (plurality of sensors; classifiers evaluating context from signals), and on dependent-claim features that appear in the specification (feature vectors from sensor readings; context-dependent actions; adaptive classifier configuration). US 2002/0128000 supplies the "identified context = in a moving vehicle" target, which the '564 claims presuppose.

Limitations NOT covered — the fatal gap. Nothing in WO or US 2002/0128000 discloses:

• hierarchically ordered sensor groups with separate classifiers per group;
• conditional sequential activation — activating a higher-level group only after the lower-level result;
• adaptation of the lower-level classifier based on the higher-level classifier's result.

WO's adaptation feedback comes from the user (explicit ± input or inferences from user behavior), not from another classifier. US 2002/0128000 has no adaptive classification at all. A PHOSITA combining these two gets an adaptive single-stage driving detector — not the cascade. Combination A would not render claims 1, 16, or 23 obvious.

Combination B — WO 2010/133770 + US 2005/0255874

What it yields. Same as Combination A, except the driving context is detected by velocity thresholding (GPS/accelerometer) or by proximity to a vehicle-mounted short-range transmitter. Again, no hierarchy, no sequential group activation, no cross-classifier adaptation. It supplies the alternative detection modalities (accelerometer, GPS receiver) that appear in the '564 dependent claims' sensor lists, but not the architecture. Not sufficient for the independent claims.

Combination C — US 2009/0128286 + US 2002/0128000 (or US 2005/0255874)

What it yields. Vehicle-side control of device usage (permit/block calls based on in-vehicle use and seat position) combined with device-side or network-side driving detection. This covers the motivation for the '564 patent's field (safe, legal use of phones in cars) but contains no adaptive classifiers, no sensor groups, no hierarchy, and no feedback-based retraining whatsoever. It is the least relevant to the claim language. Not sufficient.

Combination D — the combination the claims were actually designed to avoid

The only § 103 theory with real traction would combine WO 2010/133770 (adaptive classifier + feedback + action set) with a reference teaching hierarchical/cascaded, power-aware sensor activation ("coarse-to-fine" or "wake-up" sensing: a low-power always-on stage gates a high-power stage such as GPS), and US 2002/0128000 for the driving context. The motivation story is strong and squarely grounded in the patent's own Background:

• The '564 specification itself concedes that GPS accuracy comes at unacceptable energy cost, that base-station speed detection suffers latency and false positives, and that parallel multi-sensor analysis "substantially increase[s] the effectiveness of the classification … at the expense of increased power consumption." A PHOSITA facing this exact tradeoff would naturally look to a staged approach: cheap/imprecise sensing first, expensive/precise sensing only on trigger. Cascade and coarse-to-fine classification were well-known design patterns in pattern recognition and low-power embedded design before Feb. 2012.
• WO supplies the missing "adapt the cheap stage" mechanism in general form (on-line adaptation of classifier configuration from feedback); the leap claimed in '564 is only the source of that feedback — replacing user feedback with the higher-level classifier's result.

However — and this is critical — I could not verify, within this session's tool budget, any specific prior-art reference that teaches (i) hierarchical sensor groups with per-group classifiers, (ii) conditional sequential activation for power savings, and (iii) automated cross-level classifier adaptation. The Unified Patents bounty is telling: Unified ran a $3,000 prior-art contest specifically on claim 23 (excluding previously-cited art) and published winning art in Feb. 2024 — yet no IPR has ever been filed against this patent in the ~2.5 years since. That is consistent with the proposition that the combination needed for Combination D has not been located in citable form. I will not fabricate a reference number. The winning contest art (linked from the Unified announcement) is the single most important document to obtain before concluding the claims are vulnerable.

---

3. Motivation-to-combine analysis

For each pair above, the motivations are conventional and would be articulated by a PHOSITA as follows:

1. WO + US 2002/0128000: Both are in the same art unit — mobile-device context awareness. WO provides the general-purpose adaptive context-recognition engine; US 2002/0128000 identifies the commercially salient "driving" context and the call-management services. Motivation: apply the known adaptive-recognition framework to the known driving-detection problem to improve classification accuracy over US 2002/0128000's pure speed thresholding. This is a "known technique in one reference applied to a known problem in another" argument (KSR-type). But it still does not reach the hierarchy/sequence/adaptation limitations, so the motivation is real but insufficient.

2. Adding US 2005/0255874 or US 2009/0128286: These merely corroborate the field's focus on in-vehicle detection and call blocking; they would be cited to show the problem was recognized and that sensors like accelerometers/GPS and transmitter-proximity detection were known alternatives. They do not add the missing architecture.

3. The missing ingredient: A § 103 case on claims 1/16/23 must supply a reference (or combination) disclosing hierarchical groups of sensors with group classifiers activated in sequence, with lower-level classifiers adapted from higher-level results. My assessment: WO 2010/133770 is the closest prior art, and the '564 claims are, in substance, an obvious-in-hindsight redesign of WO's adaptation loop (swapping "user feedback" for "higher-level classifier feedback") plus a routine power-management cascade — if a suitable cascade reference exists. Without one, the claims likely survive § 103 on this record.

---

4. Dependent claims (2–15, 17–22, 24)

• Claims 2–4 (multiple classifiers per group; multi-context classifiers): Standard classifier-engineering choices; WO's per-context-class adaptation model and its multiple classification-logic settings (one per user profile) make these the most vulnerable dependent claims — but they depend from independent claims that already fail on the hierarchy/sequence limitations.
• Claims 5–6 (numerical/probabilistic outputs with threshold comparison): Expressly anticipated in kind by WO's probability-style classification and the '564's own description of thresholding; a weak limitation.
• Claims 7–10 (continuous/periodic classification, variable period dependent on location): Routine power-management scheduling. The '564 spec itself describes period adjustment based on time of day, calendar, battery state — conventional design choices.
• Claims 11 (power-state change on activating second group): This is the "power on the higher group" feature, which follows from the cascade concept; vulnerable if the cascade concept is shown in prior art.
• Claims 12–13, 15 (activation on positive result; activation on negative result; activation on low confidence): Claim 12 is the core cascade trigger; claims 13 and 15 are fallback triggers described in the '564 spec's steps 207–209. These are the heart of the invention, not routine design choices.
• Claims 14, 21–22 (definition of classification = read signals → determine features → classify): This is essentially WO's feature-determination-logic + classifier pipeline, verbatim. Obvious.

---

5. Secondary considerations

• Long-felt, unsolved need: The '564 Background describes a genuine, long-recognized tradeoff (latency vs. accuracy vs. battery life in driving detection). If the applicant can show the industry tried and failed to solve it (e.g., the failure of prior attempts to deploy practical always-on driving detection), this cuts against obviousness.
• Commercial success / copying: The patent is being asserted against GPS/navigation and automotive defendants (Garmin, rental companies, dealerships), but I have no evidence of copying or of the claimed architecture being adopted industry-wide — the litigation record shows settlements/dismissals rather than licensing of the specific cascade feature. Neutral-to-weak for the patentee.
• No IPR ever filed despite a targeted bounty: As noted, this is the strongest available signal (not evidence) that even motivated, well-funded challengers have not located art sufficient for a § 103 case on claim 23.

---

6. Bottom line

• Combinations A–C (the verified prior art) do not render claims 1, 16, or 23 obvious. They show that adaptive context classification (WO 2010/133770), driving detection (US 2002/0128000, US 2005/0255874), and in-vehicle call control (US 2009/0128286) were all known — but none teaches the specific hierarchical sensor-group cascade with downward cross-classifier adaptation that defines the independent claims.
• The only viable § 103 theory is Combination D: WO 2010/133770 + a power-aware hierarchical/cascaded sensor-activation reference + US 2002/0128000 for the driving context. The motivation (battery life vs. accuracy vs. latency) is explicitly documented in the '564 Background itself, and the cascade pattern is a routine engineering response to that tradeoff — but the theory rises or falls on locating a citable cascade reference that also (or in combination) discloses using the higher-level classifier's result to retrain the lower-level one.
• Recommended next step: retrieve the winning Unified Patents contest prior art on claim 23 (published 2024-02-14) and run targeted searches for "coarse-to-fine / cascade / hierarchical sensor activation power management," "multi-stage classifier mobile context," and "always-on low-power sensor wake GPS" references predating 2012-02-17. Those searches, which I could not complete within the tool budget, are the difference between "claims likely valid" and "claims vulnerable" on this record.

Confidence statement: High confidence on what the verified references do and do not disclose, and on the claim-element mapping above. Moderate-to-low confidence on the ultimate § 103 outcome, because the outcome depends on unverified (but plausibly existing) cascade/hierarchy prior art that I could not retrieve in this session.