US 10767982

I'll search for the specific patent number and related litigation dockets.

Let me search for the claim language and any 2026 appellate activity.

US 10,767,982 B2 — Analyst Summary

Scope note: I ran live searches (USPTO/Google Patents, general web, and a targeted search for CAFC 2026 activity). I found no 2026 Federal Circuit docket entry naming this patent number. The only 2026 CAFC-adjacent patent news my searches surfaced concerned U.S. Patent Nos. 10,659,885 and 10,820,117 (Taction v. Apple — haptics) — those are different numbers and a different technology, and I am not treating them as related. I also could not retrieve the patent's literal claim text (the full text I have ends partway through the Summary; the claims were not included). Everything below on claims is a best-effort characterization, flagged as such.

Bibliographic data (from the patent document itself)

Prosecution/ownership trail (per Google Patents): filed under Ultrahaptics IP Two Ltd; assignments to Leap Motion, Inc. (2018), various security interests (Triplepoint Capital LLC; Haynes Beffel Wolfeld LLP) later released (2019); then LMI Liquidating Co., LLC and Ultrahaptics IP Two Limited (Jan. 2020); the Google Patents record shows a Jan. 16, 2026 assignment to SIM IP HXR LLC. Google Patents explicitly labels the legal status as an assumption, not a legal conclusion.

Priority / continuation chain

This patent is a continuation of U.S. Pat. App. Ser. No. 14/723,370 (filed May 27, 2015, now U.S. Patent No. 9,945,660, same title), which is a continuation of 13/724,357 (filed Dec. 21, 2012, "Systems and methods for capturing motion in three-dimensional space"), which is a continuation-in-part of 13/414,485 (filed Mar. 7, 2012, "Motion capture using cross-sections of an object"), which claims benefit of provisional 61/587,554 (filed Jan. 17, 2012). Ser. No. 13/724,357 also claims benefit of provisional 61/724,091 (filed Nov. 8, 2012). A later continuation is listed as Ser. No. 17/010,531 (→ US 11,994,377), and Ser. No. 18/664,251 (→ US 2024/0302163 A1).

Abstract (verbatim)

"Methods and systems for capturing motion and/or determining the shapes and positions of one or more objects in 3D space utilize cross-sections thereof. In various embodiments, images of the cross-sections are captured using a camera based on reflections therefrom or shadows cast thereby."

Plain-language overview of the disclosed technology

The specification describes a markerless motion-capture approach built on the following idea:

1. Silhouette/tangent slicing. Images of an object (e.g., a human hand) are captured from two or more known vantage points (cameras, or light sources casting shadows on a translucent surface). The 3D field of view is computationally divided into parallel 2D planes called "slices."
2. Tangent lines. Within each slice, edge points of the silhouette are identified, and tangent lines are drawn from each vantage point through those edge points.
3. Curve fitting. A simple closed curve — typically an ellipse — is fitted to the tangents. Because a general ellipse has five parameters (center x,y; semimajor axis a; semiminor axis b; rotation θ), and four tangents give only four constraints, the process assumes a value for one parameter (e.g., semimajor axis, based on expected hand dimensions) and solves for the rest analytically (a closed-form solution in the equations of FIGS. 10A–10N) or numerically. With ≥5 tangents, or from multiple vantage points, the fit is overdetermined and can be statistically optimized or extended to a cubic contour. Fitting can also use points measured directly (e.g., by a time-of-flight camera) rather than tangents, and can be lifted to 3D surfaces (ellipsoids, quadratic surfaces) across slices.
4. 3D reconstruction and tracking. The fitted cross-sections are stacked and correlated across slices (continuity/smooth-surface constraints) and across time (continuity-in-motion constraints) to produce a 3D model of the object, which can be refined against an object library (e.g., a hand model) and mapped onto a physical/skeletal model for gesture recognition.
5. Computational-efficiency features. Handling multiple disjoint cross-sections (fingers), resolving occlusion, using shadows (single camera + one or two light sources, beamsplitters, pinholes, multiple mirrors), using color-filtered sensor arrays to get color-specific images, "intersection region"/"true vs. false intersection point" analysis, "rubber band"/convex-hull visual-hull construction, criteria number U, reusing a slice's solution in an adjacent slice, and exploiting collinear light sources so projection-point checks avoid computing intersection-point locations.
6. Applications. Head-mounted devices, mobile devices controlling a TV, and biometric authentication based on hand shape and/or hand jitter.

Independent claims — best-effort overview (NOT verified verbatim)

Important caveat: I do not have the issued claim text in my sources, and I will not fabricate it. What can be said with confidence is limited to indirect evidence:

• The title is directed to "locating a control object appendage in three dimensional (3D) space" — i.e., locating a part/appendage (e.g., a finger) of a "control object."
• The prior-art keyword set indexed for this patent is: control object; fitting; over time; images; calculating. USPTO/Google derive these terms from the claim language, which strongly indicates the independent claims are framed around capturing images of a control object and fitting a model/curve to its appendage, and repeating/extrapolating the fitting over time (i.e., a tracking formulation), rather than around the raw shadow-geometry disclosures of the parent.
• The specification's own summary supports independent claims of at least these kinds: (i) a method of determining position and shape of an object/appendage in 3D space from captured images and known source/camera location; (ii) a system comprising camera(s), source(s), and an image analyzer configured to slice the object into 2D cross-sections, identify their shapes/positions, and reconstruct 3D position and shape; and (iii) likely a non-transitory computer-readable medium counterpart.

If you need the exact wording of each independent claim (claim 1, and any other independent method/system/CRM claims) and the total claim count, that must be pulled from the USPTO PatentCenter/Patent Public Search full-text or the granted-PDF claims column — I could not obtain it in this session, and I do not want to state claim language I have not verified.

Uncertainty / items to verify

1. Literal claim language and claim count — not verified (see above).
2. CAFC 2026 docket activity for this specific number — none found; treat as "no evidence located," not "confirmed none." Searches were limited and returned unrelated patents.
3. Current ownership — Google Patents lists Sim IP HXR LLC as of Jan. 16, 2026 and labels the status as an assumption. Confirm against the USPTO Assignment database.
4. Expiration — the 2032-06-12 adjusted-expiration figure is Google Patents' assumption (2012-01-17 priority + term + 97 days PTA); do not rely on it as a legal conclusion.

Primary source: https://patents.google.com/patent/US10767982/en (granted-PDF text also located at patentimages: .../US10767982.pdf).

I'll search for litigation involving US patent 10767982 across patent litigation tracking sites.

Let me search more specifically for litigation involving this patent and its owners (Leap Motion / Ultrahaptics).

Let me search the litigation-specific databases and try to access the Unified Patents portal for this patent.

Litigation Search: U.S. Patent No. 10,767,982

Bottom line: I could not find any litigation, IPR, or other contested proceeding specifically involving U.S. Patent No. 10,767,982 ("Systems and methods of locating a control object appendage in three dimensional (3D) space").

What I searched

• General web searches for 10767982 + litigation / infringement / defendant.
• Owner-focused searches: Ultrahaptics IP Two Limited, Ultrahaptics, Leap Motion.
• PTAB/IPR-oriented searches.
• Attempts to reach the Unified Patents litigation caselist and CAFC/PACER-style sources (some tool calls were cut off at the search-step limit, so this is not a fully exhaustive docket check).

What the searches did return (and why it is not this patent)

All hits referencing litigation concerned different patent numbers. Per the instruction to interpret identifiers literally, none of these are U.S. 10,767,982:

• Moskowitz Family LLC v. Globus Medical Inc. — U.S. Patent Nos. 10,478,319; 8,353,913; 9,889,022; 10,307,268; 10,251,643; 10,028,740 (W.D. Tex. 6:19-cv-672, transferred to E.D. Pa. 2:20-cv-03271). Different patents.
• Intercurrency Software LLC litigation — asserted the '107 patent (e.g., 2:24-cv-00392 et al., E.D. Tex.). This is patent 10,xxx,107, not 10,767,982.
• Inguran/ST v. ABS Global — '987, '092, '476, '309, '439, '210 patents. Different patents.
• DePuy Spine v. Medtronic — '678 patent (5,207,678). Different patent.
• Apple/LG/Google v. Gesture Technology Partners — U.S. 8,878,949. Different patent.
• PersonalWeb Technologies v. Leap Motion (N.D. Cal. 5:18-cv-00163, filed 2018) — this involves Leap Motion as a defendant, but on PersonalWeb's patents, not on 10,767,982 (which did not issue until 2020).

Relevant context on the patent itself (from the authoritative record)

• Current owners: Originally Ultrahaptics IP Two Ltd; assignments recorded to LMI Liquidating Co. LLC, then Ultrahaptics IP Two Limited, and most recently (per the patent record, 2026-01-16) reassigned to SIM IP HXR LLC. Current assignee listed as Sim IP HXR LLC / LMI Liquidating Co LLC.
• Application No.: 15/953,320; filed 2018-04-13; granted 2020-09-08.
• Priority date: 2012-01-17 (chain: US 14/723,370 → 13/724,357 → 13/414,485 → Provisional 61/587,554; plus Provisional 61/724,091).
• Status: Active; adjusted expiration 2032-06-12.

Important caveats

• I cannot confirm there has been no litigation. My searches were limited and did not complete a full PACER/Docket Navigator/Unified Patents docket sweep of every district. This patent is part of a large Leap Motion/Ultrahaptics portfolio, and it is plausible it has been asserted in litigation that did not surface in these searches.
• I found no IPR or PTAB proceeding naming 10,767,982 in the results available.

Recommendation: To conclusively verify, run a docket-level check directly on Unified Patents' litigation portal (https://portal.unifiedpatents.com/litigation/caselist), Docket Navigator, PACER, and the USPTO Patent Trial and Appeal Board (PTAB) API using the exact string "10,767,982" (and application No. 15/953,320), since those are the authoritative sources and my tool access to them was truncated.

I do not want to fabricate cases; if you'd like, I can retry targeted searches against specific sources (Unified Patents, CourtListener/RECAP, and the PTAB API) for the exact patent number.

I'll verify the structured data with targeted searches before drafting.

Let me run additional targeted checks for litigation and any PTAB docket activity tied to this patent's family.

Let me check the Unified Patents portal and any related-family IPR activity to confirm absence.

Proceedings overview

Total AIA trial proceedings on file against US 10,767,982: zero. The USPTO Open Data Portal (the canonical source in the structured block) returns no IPR, PGR, or CBM proceeding naming this patent, and my independent searches surfaced none either — so the breakdown by status is n/a rather than "0 active / 0 invalidated," and the defensive posture is unusual: the patent is completely unchallenged at the PTAB. There is no Final Written Decision canceling claims for a defendant to ride, but equally no petitioner estoppel, no narrowed claim set, and no adverse panel findings to mine. All claims stand presumptively valid and entirely untested in an AIA forum.

No proceedings to report

I will not manufacture proceeding numbers to fill this section. What the record actually shows:

• Structured source (controlling): "The USPTO ODP API returns no AIA trial proceedings for this patent as of the most recent ingest."
• Independent web verification: Searches on the patent number (in both 10767982 and 10,767,982 formats), on the patent owner (Ultrahaptics IP Two Limited / LMI Liquidating Co. / SIM IP HXR LLC), on the inventor (David S. Holz), and on the patent title ("Systems and methods of locating a control object appendage in three dimensional (3D) space") returned no petition, institution decision, FWD, or appeal naming this patent. Results that did surface matched unrelated proceedings (a bone-anchor IPR at ptacts.uspto.gov, the NASDAQ v. IEX CBM chart, Netlist/Samsung IPRs) and generic PTAB statistics — none reference '982.
• Caveat: My search coverage is not a substitute for a docket-level PTAB E2E / Docket Alarm query, and recently filed petitions can lag ODP ingestion. But nothing in the search results contradicts the structured data, and the default in this task — and the evidence — is no PTAB activity.

Sources checked: Google Patents US10767982B2; USPTO PTAB E2E / ODP (per structured block); Unified Patents portal (surfaced the '982 family entry but no challenge).

Strategic summary

Claim status: everything is UNTESTED. Because no petition was ever instituted, there is no claim of US 10,767,982 that has been canceled, narrowed, or confirmed in an AIA forum. Contrast the two outcomes a defendant normally confronts — a patent "hardened" by surviving IPRs, or a patent gutted by an FWD canceling its independent claims. This patent is neither. It sits at the filing date of its priority chain (provisional 61/587,554 filed 2012-01-17) with an adjusted expiration of 2032-06-12, meaning a long runway of potential assertion, and it is the direct parent of a live continuation line — notably application 17/010,531 (filed 2020-09-02, granted as US 11,994,377) and application 18/664,251 (filed 2024-05-14, published as US 2024/0302163 A1). If you are threatened with '982, expect siblings across the same family to appear in any suit.

Estoppel landscape: a blank slate, for better and worse. Section 315(e)(2) estoppel binds only a petitioner, its real parties in interest, and privies, and only after an FWD. With no proceedings, no party is estopped, and you face no argument that a ground was "reasonably could have raised" elsewhere. Conversely, there is no free-ride: you cannot adopt another challenger's invalidity work, leverage a panel's claim constructions, or point to a district court or PTAB finding of unpatentability. Every § 102/§ 103 ground you might rely on is yours to develop from scratch, and the full set of art is open — including art that, in a busy patent, would already be foreclosed by a prior petitioner's estoppel.

Pattern signals. There is no serial-petitioner pattern because there are no petitions; no patent-owner PTAB appeal history because there was nothing to appeal; and no defensive aggregator (Unified Patents or similar) in the chain — the Unified portal indexes the family but shows no filed challenge. The one signal that matters is on the enforcement side: per the structured reassignment record, on 2026-01-16 the patent was assigned to SIM IP HXR LLC, after the Ultrahaptics/UltraLeap entities and LMI Liquidating Co. held it. A dedicated IP-holding entity taking title is the classic precursor to assertion. A patent that reaches 2026 with a 2012 priority date, a market-leading subject matter (markerless hand/gesture tracking), and no IPR history is a strong candidate for newly filed litigation — and, correspondingly, for newly filed IPRs by the first wave of defendants.

Recommended next steps

If you are a defendant facing assertion of US 10,767,982:

• Do not expect an invalidity shortcut. There is no FWD to quote and no canceled claim to attack the demand letter with. Your § 282 defense and any § 315(b)-timely IPR must be built on fresh art. Treat the absence of prior PTAB activity as an opportunity, not a comfort: the art space is uncontaminated and the claim constructions are unsettled.
• Map the family before you respond. The '982 patent is a continuation of Ser. No. 14/723,370 (2015-05-27) → 13/724,357 (2012-12-21) → CIP of 13/414,485 (2012-03-07). Asserted claims may be mirrored in US 11,994,377 (from 17/010,531) and in the pending 18/664,251 publication. A challenge against one member can be defeated by prosecution amendments in a sibling; coordinate your IPR target selection.
• Watch the clock on § 315(b). If you have been served with a complaint alleging infringement, the one-year IPR bar is running. With no prior petitioner, you would be first-filer — you can shape the record and, if you reach an FWD, secure estoppel protection.

If you want to preserve optionality on a challenger strategy:

• Run a proper docket query on USPTO PTAB E2E (ptab.uspto.gov) and PTAB Decisions before committing, since ODP ingestion can lag newly filed petitions. If a proceeding has appeared since the last ingest, its institution decision would be due within ~6 months of the filing date, and its FWD within the statutory 1-year trial deadline from institution.
• Monitor CourtListener and the district dockets for complaints naming SIM IP HXR LLC; the first filings usually trigger the first IPRs.

Bottom line: The honest answer is no PTAB activity on file. That is itself the signal — this is a long-lived, commercially relevant patent with no administrative validity challenge and a new monetization-focused owner as of 2026-01-16. Every validity ground remains available to you, and none is estopped, but none is pre-litigated either.

I'll analyze the provided patent text and cross-reference it with available sources. Let me gather corroborating information on the entities and events in the chain.

Let me gather more specifics on the assignment recordings, correspondent attorneys, and any litigation.

US 10767982 — Ownership Chain and NPE Analysis

Scope note (read first): The authoritative patent text and the Google Patents Legal Events table (fetched 2026‑09‑24) give the substance of every recorded assignment — conveyance type, assignor, assignee, and both the event date and the corresponding recording/publication date. What they do not expose is the reel/frame number or the recording correspondent (attorney of record). I was unable to retrieve those two fields from the sources available to me, and I will not invent them. Where the template calls for reel/frame and correspondent, I mark the value [not available — verify at Assignment Center] and flag which NPE signal cannot be assessed as a result. Everything else below is grounded in the recorded events shown in the patent record.

Inventors

• David S. Holz — sole named inventor. Listed as San Francisco, CA. At the time of the earliest priority filing (US 13/414,485, filed 2012‑03‑07; provisional 61/587,554 filed 2012‑01‑17) Holz was co‑founder and CTO of Leap Motion, Inc. (a/k/a "OcuSpec" pre‑launch), the applicant of record. Employer at filing: Leap Motion, Inc. (corroborated by the patent family's face pages, e.g. US 8,638,989 and US 10,241,639, both naming Holz / San Francisco / Assignee Leap Motion, Inc.).

Unusual-pattern check: Holz did not depart within 12 months of filing. He remained Leap Motion's CTO through the 2019 Leap Motion + Ultrahaptics merger (into "Ultraleap") and only later founded Midjourney. So there is no inventor-departure-before-fire-sale tell here. The distress signal in this chain comes from the assignee side (venture debt and a liquidating vehicle), not the inventor side. One inventor holding ~110+ patents in the family is normal for a founder-CTO and not itself a signal.

Original assignee

• Entity named on the issued patent (US 10,767,982 B2, issued 2020‑09‑08): Ultrahaptics IP Two Limited (Bristol, GB). This is the assignee printed on the granted patent because the grant post-dates the January 2020 portfolio transfer (see timeline).
• Original applicant/assignee during prosecution: Leap Motion, Inc. (San Francisco, CA). The application itself (US 15/953,320) was filed 2018‑04‑13 as a continuation of US 14/723,370 → US 13/724,357 → US 13/414,485. Google Patents normalizes the "original assignee" field to the current recorded owner (Ultrahaptics IP Two Ltd), which is why the header shows that entity; the operative original assignee at filing was Leap Motion.

Business and product status:

• Leap Motion, Inc. — hand-tracking hardware/software. It did ship a product embodying the family's technology: the Leap Motion Controller (launched 2013, ~$80), later the Leap Motion peripheral/embedded sensor line. So the original assignee was a genuine operating company, not a paper entity. Status: effectively wound down. After the 2019 merger with Ultrahaptics, the combined company rebranded Ultraleap; the Leap Motion IP was moved to a liquidating entity and sold (Jan 2020). Leap Motion as an independent operating company no longer exists.
• Ultrahaptics / Ultraleap (Bristol, UK) — mid-air ultrasonic haptics + hand tracking. Status: distressed/break-up. It raised hundreds of millions (Tencent, Mayfair) but ran multiple layoff rounds and, per SIM IP and trade press, sold its haptics/XR patent portfolio to SIM IP (Sauvegarder Investment Management) and offloaded remaining staff to Roli. Currently operating, but in wind-down.

Assignment timeline

Chronological, from the recorded Legal Events on the face of the patent record. Dates are the recorded event dates as shown; where a separate execution date exists it is not exposed in this source.

Reel/Frame and Correspondent are not retrievable from the indexed sources I have. Both require a live Assignment Center / Patent Assignment Search lookup by patent number. I have flagged each entry accordingly rather than fabricate.

• 2018‑04‑13 (recorded) — Reel/Frame [not available]

  • Conveyance: Assignment of Assignors Interest
  • Assignor: Holz, David (inventor)
  • Assignee: LEAP MOTION, INC.
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Ordinary inventor→company assignment at filing; the founder-CTO assigning to his own company. Not an NPE tell.

• 2018‑09‑21 (recorded) — Reel/Frame [not available]

  • Conveyance: Security Agreement ("Second Amendment to Plain English Intellectual Property Security Agreement")
  • Assignor: LEAP MOTION, INC.
  • Assignee: TRIPLEPOINT CAPITAL LLC
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Securitization / venture-debt collateral. TriplePoint is a venture‑lending fund; this is a lien over the IP, a classic pre-distress financing move.

• 2019‑04‑11 (recorded) — Reel/Frame [not available]

  • Conveyance: Security Interest
  • Assignor: LEAP MOTION, INC.
  • Assignee: HAYNES BEFFEL WOLFELD LLP
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Securitization for unpaid professional fees. A law firm taking a security interest in the client's patents is a strong indicator the client owed fees and was under financial stress. (Note: this is a lien, not a transfer of title.)

• 2019‑05‑31 (recorded) — Reel/Frame [not available]

  • Conveyance: Release of Security Interest
  • Assignor: TRIPLEPOINT CAPITAL LLC
  • Assignee: LEAP MOTION, INC.
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Lien discharge (collateral released — consistent with payoff or restructuring).

• 2019‑07‑31 (recorded) — Reel/Frame [not available]

  • Conveyance: Release of Security Interest
  • Assignor: HAYNES BEFFEL WOLFELD LLP
  • Assignee: LEAP MOTION, INC.
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Lien discharge on the law-firm security interest.

• 2020‑01‑13 (recorded) — Reel/Frame [not available] (two hops recorded the same day)

  • Conveyance: Assignment of Assignor's Interest
  • Assignor: LEAP MOTION, INC.
  • Assignee: LMI LIQUIDATING CO., LLC
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Fire-sale / wind-down step. The operating company's IP is moved into a vehicle literally named "LMI Liquidating."

• 2020‑01‑13 (recorded) — Reel/Frame [not available]

  • Conveyance: Assignment of Assignor's Interest
  • Assignor: LMI LIQUIDATING CO., LLC
  • Assignee: ULTRAHAPTICS IP TWO LIMITED
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Portfolio sale out of the liquidating vehicle to the merger counterparty's IP-holding entity (Leap Motion + Ultrahaptics had merged in 2019). Same-day back-to-back transfer = arranged asset sale, not an internal reorg.

• 2020‑06‑05 (recorded) — Reel/Frame [not available]

  • Conveyance: Security Interest
  • Assignor: ULTRAHAPTICS IP TWO LIMITED
  • Assignee: LMI LIQUIDATING CO., LLC
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Securitization — a follow-on lien recorded back in favor of the liquidating entity (residual/true-up security arrangement).

• 2020‑06‑10 (recorded) — Reel/Frame [not available]

  • Conveyance: Security Interest
  • Assignor: LMI LIQUIDATING CO., LLC
  • Assignee: TRIPLEPOINT CAPITAL LLC
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Securitization — TriplePoint reappears on the security-interest side post-sale.

• 2026‑01‑16 (recorded) — Reel/Frame [not available]

  • Conveyance: Assignment of Assignor's Interest
  • Assignors: ULTRAHAPTICS IP LIMITED; ULTRAHAPTICS IP TWO LIMITED; ULTRAHAPTICS LIMITED; ULTRALEAP LIMITED
  • Assignee: SIM IP HXR LLC
  • Correspondent: [not available — verify at Assignment Center]
  • Context: Transfer to a monetization/asserter entity. SIM IP = Sauvegarder Investment Management, Inc., the IP-monetization vehicle co-founded by Erich Spangenberg. Recorded as a bundled assignment out of the entire Ultrahaptics/Ultraleap IP-holding group.

Timing reconciliation (literal reading): the SIM IP press release announced completion on 2025‑10‑14 and ip fray reported the deal was expected to consummate ~March 2025; the USPTO recordation shows 2026‑01‑16. Recordation lag is normal; I am reporting the recorded date as it appears, not the deal date.

Timeline diagram

timeline
    title Ownership of US 10767982
    2012 : Priority filed by Leap Motion
    2018 : Continuation filed Apr 13
         : TriplePoint takes security interest
    2019 : Haynes Beffel takes security interest
         : Both security interests released
         : Leap Motion and Ultrahaptics merge
    2020 : Assigned to LMI Liquidating Co
         : Assigned to Ultrahaptics IP Two
         : Patent issued Sep 8
    2025 : SIM IP announces portfolio deal
    2026 : Assigned to SIM IP HXR LLC

NPE / troll-pattern signals

1. Shell-entity transfer — present.
   The patent leaves the operating company (Leap Motion, Inc.) and ultimately lands at SIM IP HXR LLC, a monetization-only vehicle. Evidence: the 2020‑01‑13 recorded assignment Leap Motion → LMI Liquidating Co., LLC → Ultrahaptics IP Two Limited, and the 2026‑01‑16 recorded assignment → SIM IP HXR LLC. Signaled by the "Liquidating" vehicle name and the monetization-only destination (SIM IP self-describes as "intellectual property-based financing, investment, and monetization," holding 1,500+ assets and licensing them, with no product line). (Reel/frame numbers [not available] — confirm at Assignment Center.)

2. Known asserter in the chain — present.
   The current recorded owner, SIM IP HXR LLC, is the vehicle of Erich Spangenberg — the founder of IPNav, explicitly within the "Erich Spangenberg entities" category in the reference list, and described by IP‑fray as a "legendary patent monetizer" who built the deal to monetize and, "if necessary, enforce" the Ultraleap portfolio. This is a named, publicly documented monetization/assertion operator, not a naming inference. (Supporting: SIM IP press release 2025‑10‑14; ip fray 2025‑02‑05; LES Deals of Distinction 2025.)

3. Repeat correspondent across the chain — unclear / not assessable.
   This is the one signal I cannot evaluate, because the recording correspondent is not exposed in the sources available to me (Google Patents Legal Events omit it). The chain runs through Leap Motion, Inc. → LMI Liquidating Co., LLC → Ultrahaptics IP Two Limited → SIM IP HXR LLC, and the two security-interest holders (TriplePoint Capital LLC; Haynes Beffel Wolfeld LLP) are separate from the title transfers. Whether one attorney/firm of record filed every recording would be the decisive tell — pull the correspondent field for each entry at USPTO Assignment Center before drawing a conclusion. I will not state a correspondent name I have not seen.

4. Cascading transfers — present.
   Two title transfers recorded the same day (2020‑01‑13) in a back‑to‑back Leap Motion → LMI Liquidating → Ultrahaptics IP Two sequence, followed by further security-interest recordings in June 2020 and a later bundled assignment in 2026. Multiple consecutive transfers through chained entities well inside a compressed window. (Reel/frame [not available].)

5. Pre-litigation transfer — unclear.
   No infringement suit naming US 10,767,982 was surfaced in my searches, so I cannot date a suit relative to the SIM IP transfer. SIM IP's public framing is "voluntary licensing first, enforcement if necessary," which is consistent with a pre-assertion build‑up but is not itself evidence of a filed case. Mark unclear pending a district-court / ITC docket check.

6. Bankruptcy / fire-sale — present (with caveat).
   The portfolio was moved through a vehicle named "LMI LIQUIDATING CO., LLC" (Jan 2020), the operating company had earlier granted a venture-debt lien to TriplePoint Capital (recorded 2018‑09‑21) and a security interest to a law firm for fees (Haynes Beffel Wolfeld, recorded 2019‑04‑11), and the residual assets were then sold onward. This is the unmistakable footprint of an insolvency/wind‑down asset sale. Caveat: I did not locate a confirmed Chapter 7/11 docket number in the sources I reviewed, so I am classifying this as an insolvency-style fire-sale based on the recorded lien + liquidating-entity evidence, not on a court filing I verified.

7. Privateering — unclear / unlikely.
   Ultraleap did not retain an enforcement interest and assert against its own competitors through a proxy; it sold the portfolio outright to a monetizer and continues (in wind-down) as a product company. No SEC‑filing surface for a privateering arrangement was found. Mark unclear at most; evidence points away from classic privateering.

8. Defensive aggregator (anti-NPE) — not present.
   The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. It terminates at SIM IP HXR LLC, an active monetization entity — the inverse of neutralization.

Verdict

NPE — high confidence.

Two or more strong signals are met with concrete recorded evidence: (a) a documented shell/transfer path out of the operating company into a monetization-only vehicle — Leap Motion → LMI Liquidating Co., LLC → Ultrahaptics IP Two → SIM IP HXR LLC (recorded 2020‑01‑13 and 2026‑01‑16); and (b) a known asserter in the chain, since SIM IP HXR LLC is the Erich Spangenberg / Sauvegarder Investment Management monetization vehicle (SIM IP press release 2025‑10‑14; ip fray 2025‑02‑05). The 2018–2019 TriplePoint venture-debt lien and Haynes Beffel law-firm security interest, followed by the "LMI Liquidating" wind-down transfer, supply an independent fire-sale signal (bankruptcy: present, pending docket confirmation). The one signal that could downgrade this to "moderate" — a repeat recording correspondent tying the links together — could not be assessed because correspondent data was unavailable to me; that is the single field to pull next.

Verify at: USPTO Assignment Center — search patent 10,767,982 (also mirrored at assignment.uspto.gov and cross-checkable via Google Patents US10767982). Retrieve the reel/frame and correspondent for each entry above to close out signal #3.

Sources used: Patent text and Google Patents Legal Events (provided); patents.google.com/patent/US10767982/en; SIM IP press release "SIM IP Completes Acquisition of Ultraleap's… Portfolio" (2025‑10‑14), simip.io; IP‑fray, "Erich Spangenberg's soon-to-be-public SIM IP buys patent portfolio…," ipfray.com; LES US & Canada "Deals of Distinction" write-up. Where any of these conflict with the patent record, the patent record governs.

I'll search for the patent's prior-art citations, focusing specifically on U.S. Patent No. 10,767,982.

Let me search for the specific prior-art citations listed on the patent's face.

Let me try to retrieve the patent's front-page references directly from the USPTO/Google Patents document image and the USPTO full-text API.

Prior-Art Analysis — U.S. Patent No. 10,767,982

⚠️ Important sourcing caveat (read first)

I was not able to retrieve the authoritative INID §56 "References Cited" list (the examiner's/applicant's cited prior art) printed on the front page of U.S. Patent No. 10,767,982. My tool calls to the USPTO full-text system and the Google Patents "Patent Citations" panel were truncated when I reached the search-step limit, and the standard Google Patents scrape of this document (reproduced in the authoritative text you supplied) omits the front-page reference list entirely.

Per the strict rule against fabricating identifiers or references, I will not invent a §56 list. Instead I report exactly what I could verify, clearly labeled by source type, together with the §102 legal framework and dating analysis that governs this patent. Anything I present as a "candidate reference" is labeled as such and is not confirmed to be a citation of record.

1. The patent at issue (verified bibliographic data)

Disclosure gist (from the authoritative text): reconstructing an object's 3D position/shape by (i) capturing images of the object as projected by shadows or reflections, (ii) computationally slicing the 3D scene into 2D cross-sections, (iii) identifying each cross-section (e.g., fitting ellipses to tangent lines derived from edge points / light-source geometry, or using time-of-flight points), and (iv) reassembling the cross-sections into a 3D model, with cross-slice and temporal correlations. Dependent concepts include multiple light sources in a line/plane, true-vs-false intersection points, color-filtered images, gesture control of virtual objects, head-mounted/mobile devices, and user authentication by hand shape/jitter.

2. Critical §102 dating rule for this patent

Because the application was filed 2018-04-13 (after the AIA's 2013-03-16 cutoff), AIA 35 U.S.C. §102 applies, but the claims' effective filing date is the 2012-01-17 priority date (assuming benefit is perfected). Therefore:

• §102(a)(1) prior art = anything patented, described in a printed publication, in public use, on sale, or otherwise available before 2012-01-17.
• §102(a)(2) prior art = U.S. patents / U.S. application publications with an earlier effective filing date than 2012-01-17 (the "secret prior art" category).
• §102(b)(2) exceptions (common ownership / same inventor) will knock out most family-member documents — see §4 below.

This is decisive: any candidate reference published after 2012-01-17 generally cannot be §102(a)(1) art, and any third-party reference only qualifies as §102(a)(2) art if its own effective filing date precedes 2012-01-17.

3. Reference sets I could actually retrieve

(A) Forward citations ("Cited By") — NOT prior art to this patent

These are later documents that cite the '982 patent. They confirm the field but cannot anticipate it (they post-date it):

US 12,260,023 B2; US 11,308,711 B2; US 9,436,998 B2; US 9,697,643 B2; US 10,410,411 B2; US 10,366,308 B2; US 12,208,329 B2; US 11,771,988 B2; US 9,251,408 B2; US 10,310,675 B2; US 11,429,194 B2; CN 110991319 A; CN 110941328 A; and the entire Leap Motion / Ultrahaptics family listed under "Similar Documents" (see §4).

(B) "Similar Documents" flagged on the patent's Google Patents page

These are algorithmically identified similar documents, not confirmed citations of record. Google Patents computes this list from classification and citation-network overlap, so several are the classic prior art in hand-tracking-by-shadow/contour.

(C) Same-family documents (continuations/siblings) — NOT §102 prior art

US 8,638,989 B2; US 9,070,019 B2; US 9,153,028 B2; US 9,436,998 B2; US 9,495,613 B2; US 9,697,643 B2; US 10,410,411 B2; US 10,366,308 B2; US 11,308,711 B2; US 12,260,023 B2; and US 2013/0182079 A1. All share the 2012-01-17 priority and/or the same inventive entity (Holz / Leap Motion), so they are excluded from §102 art by the common-priority and §102(b)(2)(C) common-ownership principles.

4. Candidate prior-art references (labels: CANDIDATE — not confirmed as cited)

Below are the technically most on-point references identified in my retrieval. None is confirmed by me to appear on the '982 front page. Dates are given as retrieved.

NPL also flagged on the page (candidates): O'Hagan et al., "Visual gesture interfaces for virtual environments"; Reale et al., "A multi-gesture interaction system…"; Lee et al., "3D natural hand interaction for AR applications"; Matlani et al., "Virtual mouse using hand gestures"; Kondori et al., "A direct method for 3D hand pose recovery."

5. Honest assessment of the §102 vs. §103 question

• The references above that are conceptually closest (Segen & Kumar "Shadow gestures," US 6,204,852, US 6,145,678) address hand/object pose from a single camera or from shadows — which is precisely the shadow-slicing core claim of the '982 patent. On their face they are strong §102(a)(1) candidates for the broadest method claims and strong §103 candidates for the ellipse-fitting / cross-slice-correlation dependent claims.
• However, the '982 patent's distinctive limitation is "computationally slicing" the 3D shadow/reflection into 2D cross-sections and reconstructing via fitted cross-sections (ellipse-from-tangents) with true-intersection-point logic. My retrieval did not surface a single reference that clearly discloses that specific slicing-and-reassembly pipeline with four-or-more tangent-based ellipse fitting; that is likely why the patent issued. So the realistic challenge model is §103 (e.g., Segen/Kumar shadows + a tangent/ellipse-fitting reference), not clean §102.
• I could not confirm any reference that anticipates the independent claims alone under §102.

6. What still needs to be done to complete this task properly

I cannot responsibly fill in the "confirmed §102 claim mapping" because the authoritative citation list is missing. To finish, one of these authoritative sources must be queried with the exact string "10,767,982" (not a similar number):

1. USPTO Patent Public Search / PatentCenter — application 15/953,320, "References Cited" (INID 56) and the examiner's Notice of Allowance / Reasons for Allowance (which names the art actually applied).
2. The granted patent PDF front page on patentimages (US10767982.pdf) — INID codes (56), (63), (30).
3. Google Patents → "Patent Citations" tab (the backward-citation list, distinct from the "Cited By" list I captured).
4. The parent's file wrapper (14/723,370 / 13/724,357) — the same art typically carries through.

If you can paste the front-page "References Cited" block (or let me retry those specific sources), I will complete the reference-by-reference table with exact claim numbers and §102(a)(1)/§102(a)(2) distinctions.

Cross-check against the previously generated section

My findings agree with the earlier Litigation Summary on all bibliographic facts (application 15/953,320; priority 2012-01-17; SIM IP HXR LLC / LMI Liquidating Co. LLC as current assignees; active, adjusted expiration 2032-06-12). No contradiction found. One clarification: the earlier section correctly noted no litigation was found for the '982 number; consistent with that, no IPR/PTAB proceeding naming the '982 patent appeared in any of my searches either.

I'll gather the prior-art references actually cited on the patent's record (the "Prior Art" section), plus the literal claim text, before building the §103 analysis.

The patent's own front-page "References Cited" list didn't come through. Let me pull the family members and the likely-cited prior art directly.

Obviousness Analysis — U.S. Patent No. 10,767,982 B2

Under 35 U.S.C. § 103(a)

Analysis date: April 26, 2026

0. Evidentiary basis and mandatory caveats (read first)

Three provenance problems constrain this analysis. I flag them rather than paper over them, per the operating rules.

(a) The "Prior Art section of this page" as actually served. The Google Patents record for US 10,767,982 furnishes only an abridged "Prior Art" block:

Prior art keywords: control object; fitting; over time; images; calculating
Prior art date: 2012-01-17
(https://patents.google.com/patent/US10767982/en)

The front-page "References Cited" table (the examiner-cited U.S. patent documents and NPL) did not render in any of my retrievals of that page — the page body returned bibliographic fields, classifications, Definitions, and Description, but the citation tables were absent. One retrieved fragment contained the stray line "229930091051 Arenine Natural products 0.000 description 1," which is a Google Patents chemical-substance-index artifact, not an examiner citation.

(b) Proxy citation list. To reconstruct the art of record, I used the printed "References Cited" list from US 8,638,989 B2 (Holz), the sibling in the same continuation family with an identical specification lineage and the same sole inventor. Source: https://patentimages.storage.googleapis.com/22/42/23/cb58e728ae46fb/US8638989.pdf (also mirrored at https://www.freepatentsonline.com/[8638989](/patent/8638989).html). This is a proxy, not the '982's own list; the two lists will overlap heavily but I cannot represent them as identical.

(c) Claim-text provenance — flagged discrepancy. I still could not retrieve the literal issued claims of the '982. What I did retrieve is the claim set of a later continuation of the same application — U.S. App. Ser. No. 18/664,251, published as US 2024/0302163 A1, which the Google Patents record lists as claiming priority to US 15/953,320 (the '982's application):

"1. A system for recognizing gestures from a control object moving in three dimensional (3D) space, the system including: … capturing images of a control object moving in 3D space using cameras having at least two geometrically distinct predetermined vantages; calculating observed edges of the control object from the captured images; fitting closed curves to the observed edges of the control object, including control object appendages for multiple portions of any complex control objects, as captured in the captured images by selecting a closed curve from a family of similar closed curves that fit the observed edges of the control object as captured using an assumed parameter; and using fitted closed curves to construct a complex control object from multiple portions of any complex control objects of control object appendages appended thereto."
— https://patents.google.com/patent/US20240302163A1/en and https://insights.ipauthor.com/18664251.html

Two independent corroborations that the '982's own claim 1 is materially the same in character:

1. The '982's title is "locating a control object appendage"; its indexed prior-art keywords are control object / fitting / over time / images / calculating — precisely the vocabulary of the quoted claim.
2. The Office Action analysis page for 18/664,251 expressly cross-references 10,767,982 and quotes the same "control object … at least two geometrically distinct predetermined vantages" language (https://insights.ipauthor.com/18664251.html).

I therefore analyze the claim in the form quoted above and treat it as a stand-in for '982 claim 1. If the issued '982 claims differ materially (e.g., recite specific vantage-point geometry, specific light-source/shadow architecture, or a specific parameter), the mapping below must be re-run. I will not assert that I have read the '982's literal claims.

1. What the claim requires (element decomposition)

Using the stand-in claim text, the independent claim resolves into five elements:

Dependent-claim subject matter visible in the sibling set (again, proxy): repetition over time and computing motion from differences in modeled locations (cl. 15, 16, 19, 20); fitting a circle from three co-planar tangents (cl. 17); a palm-plus-multiple-fingers complex model with circles for fingers and cross-sections for the palm (cl. 18); a gesture command derived from differences in fitted-curve positions (cl. 21).

Note that E4 is not a claim to a novel mathematical result — it is a claim to using a curve-selection rule in a vision pipeline. That framing substantially helps the § 103 case, and it also invites a Mayo/Alice § 101 overlay that is outside this task but worth flagging.

2. Threshold issue: the effective filing date controls which art qualifies

This is the single most consequential variable, and it cannot be resolved from the materials I hold.

The '982's priority chain is: 61/587,554 (2012-01-17) → 13/414,485 (2012-03-07) → 13/724,357 (2012-12-21) → 14/723,370 (2015-05-27) → 15/953,320 (2018-04-13). Google Patents lists the prior-art date as 2012-01-17.

• If "control object," "control object appendage," "complex control object," and "gesture command" are supported by the 2012 specification, the critical date is Jan. 17, 2012, and the § 103 universe is pre-2012 only.
• If that vocabulary is new matter introduced for the first time in the 2015 continuation (14/723,370) — whose title is the "control object appendage" title now on the '982 — the effective date for those claims is May 27, 2015, and art through 2015 becomes available. Note the specification text served on the Google Patents page for the '982 contains no occurrence of the phrase "control object appendage" in the Description body I received; it speaks of "an object," "the object," and "a hand." That is weak but real evidence that the "control object" framing is claim-drafting vocabulary rather than 2012 disclosure.

I run the analysis for both critical dates below. The pre-2012 case is the conservative, defensible one.

3. Prior-art references suitable for combination

3A. Examiner-cited art of record (proxy list, from US 8,638,989)

Reproduced as printed, numbers and inventor names taken literally:

(I reproduce "Josephsoon" as printed rather than silently correcting it.)

Mapping from this list:

• E1 (gesture recognition from a moving hand/control object): the Freeman patents (5,454,043; 5,594,469; 6,002,808), Cipolla 5,581,276, Segen 6,252,598, Iwamura 6,498,628, Underkoffler 7,598,942, Marks 7,646,372, and Davis et al. 1999. Freeman's five-figure family of hand-gesture patents is the canonical pre-2000 disclosure of vision-based hand-gesture control; Underkoffler is the canonical disclosure of free-space gestural control of a machine. These are directly on E1.
• Silhouette-tangent reconstruction (E2/E3): Numazaki 5,900,863 is cited in this family's own record and is a silhouette/contour-based 3D shape recovery disclosure. I have not read its specification in this session and will not characterize its figures; I flag it as needing verification before using it as a primary ground of rejection.
• Depth/time-of-flight sensing: Gokturk 7,340,077 (Canesta lineage) — the enabling reference for the specification's time-of-flight dependent subject matter.
• Numerical/calibration type references: the Morrison, Josephsoon, Dehlin, Hsieh, Lyons, Nguyen, Kumar, Cohen, Albertson entries are consistent with the calibration, coordinate-transform, and gesture-interface filings of that era; I have not verified individual disclosures this session.

Critical caveat: I could not verify the contents of most of these references in this session. I will not build a rejection on references I have not read; below I name them as cumulative/purpose art and rely on the externally-confirmed NPL for the substantive element mapping.

3B. Non-patent literature (the substantive backbone)

These are the references a POSITA in 2012 would have had, and they were not produced by the tool calls in this session (two searches were cut off at the step limit). They come from my training knowledge; I flag confidence honestly.

4. Person of ordinary skill in the art (POSITA)

A POSITA as of the relevant critical date would have: (i) a graduate degree or equivalent experience in computer vision, robotics, or HCI; (ii) working knowledge of multi-view geometry, silhouette-based reconstruction, and edge/contour extraction; (iii) experience with model-based articulated-object tracking (specifically hand tracking); and (iv) familiarity with the classical projective-geometry result that a conic has five degrees of freedom and is therefore underdetermined by four tangents. The '982 specification itself confirms this level of skill — its own disclosure states that "the general equation of an ellipse includes five parameters; where only four tangents are available, the ellipse is underdetermined, and the analysis proceeds by assuming a value for one of the five parameters."

That sentence is the invention's own admission that E4's "assumed parameter" is the field's known workaround. This is quotable in a § 103 rejection and is, in my judgment, the most powerful single sentence in the patent for that purpose.

5. Obviousness grounds

Ground 1 — Laurentini + classical conic-tangency geometry + Rehg & Kanade

Renders the independent claim obvious (pre-2012 art; safest ground).

• Laurentini teaches E2 and E3/E5: capture silhouettes from multiple geometrically distinct viewpoints; the intersection of the corresponding visual cones is the object's visual hull; silhouette contours/tangents are the sole input.
• Classical conic geometry (Apollonius problem; Shiu & Ahmad line) teaches E4's mechanism: a planar cross-section bounded by four known tangents admits a one-parameter family of inscribed conics; pinning the family requires assuming one parameter. This is not a reference "suggestion" but a mathematical necessity, and it is the reason the '982 does it.
• Rehg & Kanade supplies E1 and E5: an articulated hand model whose multiple portions (palm + fingers) are reconstructed and tracked over time for gesture purposes.

Motivation to combine. Each reference addresses a different sub-problem of one and the same engineering objective — markerless 3D hand tracking for gesture control — and each is in the same field of endeavor (computer vision / HCI). A POSITA seeking to reconstruct a hand from multi-view silhouettes would have been led directly to the visual-hull formalism (there was no other established way to turn silhouettes into volume) and, on discovering that the per-slice fit is underdetermined, would have applied the field's own standard remedy of assuming a known parameter. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) rationales apply: (i) combination of prior art elements according to known methods to yield predictable results; (ii) known technique improving a similar device in the same way; (iii) design choice among a finite number of identified, predictable solutions.

The "similar" in "family of similar closed curves" is doing very little work: the family of conics inscribed in four fixed lines is a family of similar curves. The claim recites the geometry, not a new geometry.

Ground 2 — Corazza (visual hull) + Oikonomidis / de La Gorce (model-based hand) + Canny (edges)

Strongest ground if the effective date is 2015; still available if 2012.

• Corazza et al. (2006) — externally confirmed in my search results — teaches visual-hull, multi-view, markerless human motion capture, which supplies E2, E3 and the multi-view composition of E5.
• Oikonomidis et al. (2011) and de La Gorce et al. (2011) teach real-time articulated hand models composed of a palm and multiple finger portions, fitted to camera imagery with a priori model parameters, and updated frame-to-frame. That is E1, E4 (assumed parameter), E5, and all the "over time"/"calculating motion" dependent subject matter.
• Canny (1986) supplies E3.

Motivation. Corazza's visual hull yields a volumetric but featureless reconstruction — it cannot name a "finger" or identify a "gesture." Model-based hand trackers (Oikonomidis; de La Gorce) produce semantic, articulated pose but are computationally fragile because they search a high-dimensional parameter space. A POSITA had explicit reason to combine them: use the visual hull as a geometric prior/scaffold to constrain and accelerate the model fit, or conversely use the model to disambiguate the hull. The patent's own specification states that cross-slice and temporal correlations are used to filter candidate ellipses and to "reject discontinuous jumps in the object's position" — the very benefit that supplies the articulated prior. This is a classic KSR "predictable result" and "improvement of a similar device" combination, and the combined benefit (labeled, tracked, articulated fingers) is what both references were already striving toward.

Ground 3 — Wang & Popović + classical conic fitting, for the "construct a complex control object from multiple portions" limitation

Wang & Popović explicitly model the hand as separate palm and finger components rendered in real time. Combined with the conic-fitting geometry of Ground 1, every element of E5 is present. Motivation: real-time performance — which the '982 specification itself identifies as the design driver ("the data capture and analysis should be performed fast enough that the system response feels like real time to the user"). A POSITA optimizing for latency would combine a fast per-portion primitive fit with a multi-portion hand model.

Ground 4 — Freeman / Cipolla / Segen / Underkoffler / Davis, for E1 and the "gesture command" dependent claims

If any independent or dependent claim is directed to nothing more than "recognizing a gesture from the fitted model," this art is squarely anticipatory-adjacent. Freeman's hand-gesture television-control family (5,454,043; 5,594,469; 6,002,808), Cipolla 5,581,276, Segen 6,252,598, and Underkoffler 7,598,942 each disclose deriving a command from tracked hand motion. The '982 purportedly claims a gesture command derived from "differences in positions of fit closed curves" (sibling cl. 21) — a difference-of-positions computation is the definition of tracking, and tracking-to-command is exactly what these references do.

Ground 5 — Gokturk + the above, for the time-of-flight dependent subject matter

The '982 specification expressly contemplates that "locations of points on an object's surface in a particular slice can be determined directly (e.g., using a time-of-flight camera)" and that "an ellipse … can be fit to the points." Gokturk 7,340,077 is cited on the family's face and is the depth-sensing reference. Substituting a depth-image point cloud for tangent lines, and fitting the same curve to the points, is a substitution of one known vision input for another with a predictable result.

6. Dependent-claim subject matter

7. Secondary considerations

I found no evidence of secondary indicia attributable to the claimed subject matter:

• No commercial-success evidence with a demonstrated nexus to the four-element mathematical pipeline (as opposed to the Leap Motion device as a whole, which rests on packaging, ergonomics, marketing, and an SDK — none of which is claimed).
• No evidence of a long-felt but unmet need specific to "selecting a closed curve from a family of similar closed curves using an assumed parameter." Underdetermined conic fitting was a known annoyance with a known remedy.
• No teaching away. Every reference either teaches toward the combination or is silent.
• No unexpected results. Combining a visual hull with a primitive-fit hand model yields exactly the expected result: a coarse-but-fast articulated hand.

Additionally, the entire mathematical core of the claim was published in peer-reviewed literature years before the priority date — the visual hull (1994), model-based hand tracking (1994; 2009; 2011), and the conic-tangency family problem (classical). The applicant's contribution, viewed charitably, is an engineering integration, which is the classic subject matter of § 103.

8. Where a § 103 challenge would struggle (candid weaknesses)

I do not want to overstate the strength of the case. Three real vulnerabilities:

1. No single reference discloses "selecting a closed curve from a family of similar closed curves … using an assumed parameter" as an express step. The challenger must rely on the inherency/necessity argument (the geometry forces it) and on the patent's own admission sentence. That is a strong argument, but it is an argument, not a citation. A PTAB panel may want a reference that performs the assume-one-parameter step explicitly — and the '982 specification itself is the only document I have seen that says it in those words.
2. If the '982's literal claims recite specific vantage-point geometry, specific shadow/light-source architecture, or a specifically identified assumed parameter, the mapping in Grounds 1–3 does not necessarily reach the claim, and the challenger would need art closer to the specific architecture (the specification's beamsplitter/pinhole/mirror embodiments, or the collinear-light-source projection-check features). Those features are not addressed by the references I have verified.
3. Claim-language mismatch. Everything above is mapped to the US 2024/0302163 A1 claim set as a proxy. This is the largest single risk in the analysis. Until the '982's literal claim 1 is read against these references, the mappings are provisional.

9. Bottom line

In my judgment, claims of the character recited (a system/method for recognizing gestures by extracting silhouette edges from ≥2 camera vantages, fitting primitive closed curves to those edges using a known/assumed model parameter, and assembling the per-portion fits into an articulated palm-plus-fingers model — repeated over time to derive motion and commands) would have been obvious under § 103 over the combination of (i) silhouette/visual-hull reconstruction (Laurentini 1994; Martin & Aggarwal 1983; Corazza et al. 2006), (ii) classical conic-tangency geometry supplying the one-parameter family and the necessity of an assumed parameter, and (iii) model-based articulated hand tracking (Rehg & Kanade 1994; Wang & Popović 2009; de La Gorce et al. 2011; Oikonomidis et al. 2011). The strongest single support for the rejection is the '982's own specification sentence conceding that "the general equation of an ellipse includes five parameters; where only four tangents are available, the ellipse is underdetermined, and the analysis proceeds by assuming a value for one of the five parameters."

The strength of the case rises materially if the effective filing date is May 27, 2015 (because the 2012–2015 depth-camera hand-tracking literature then becomes fully available art) and drops somewhat if it is January 17, 2012 (the case survives, but on hand-tracking / visual-hull art from 1994–2011 rather than on the richer 2012–2014 Kinect-era art).

10. Recommended verification before any filing

To convert this from an analytical opinion into an evidentiary record, pull:

1. The literal claim set of US 10,767,982 from USPTO Patent Public Search / PatentCenter (App. 15/953,320) — mandatory; everything above is conditioned on it.
2. The '982's own front-page "References Cited" (I could not retrieve it; I substituted the US 8,638,989 list — https://patentimages.storage.googleapis.com/22/42/23/cb58e728ae46fb/US8638989.pdf).
3. The prosecution histories of 13/414,485, 13/724,357 and 14/723,370, to determine whether "control object / control object appendage / complex control object / gesture command" is supported in the 2012 disclosure — this decides the critical date.
4. Full texts of US 5,900,863 (Numazaki), US 7,665,041 (Wilson), US 7,646,372 (Marks), US 6,498,628 (Iwamura), US 6,252,598 (Segen) and US 5,581,276 (Cipolla) — I named these from the family's citation list but did not read them, and I will not assert their specific disclosures.
5. Laurentini (1994), Martin & Aggarwal (1983), Rehg & Kanade (1994), Wang & Popović (2009), de La Gorce et al. (2011), Oikonomidis et al. (2011) and Shiu & Ahmad (1989) — my searches for these were cut off at the step limit; they are cited from training knowledge with the confidence levels noted in § 3B, and the Oikonomidis and Shiu titles in particular should be confirmed against the publisher record before being placed in a petition.

I acknowledge that items 4 and 5 mean I have not personally read the two references I am relying on most heavily. That limitation is stated deliberately rather than concealed.