Showcase engagement — Cortex Robotics is a fictional company. The inventors named below, their employment histories, and the ownership questions raised about them are fabricated for this demonstration. Real companies are named inside that fabricated scenario (for example, as a fictional inventor’s former employer): nothing here describes any real company’s actual work, agreements, or legal position, and none of them is affiliated with, or has endorsed, this platform or this demonstration.
Event-camera + LiDAR sensor fusion for sub-100ms obstacle detection
A sensor-fusion system for mobile robots operating in mixed warehouse environments, comprising a high-speed event camera [src-1], a solid-state LiDAR depth unit [src-2], and a purpose-built alignment kernel that reconciles the two fundamentally dissimilar data streams into a coherent obstacle field at an effective rate of 240Hz.
The novelty does not reside in any one sensor. Event cameras and solid-state LiDAR are each commercially mature. It resides in the alignment kernel — a hardware-software boundary that timestamps pixel events against discrete LiDAR frames to within 8ms [src-3][founder intake Q7], and an accompanying reactive controller that consumes the fused field directly, without a frame-based intermediate representation.
The practical result is end-to-end sensor-photon-to-avoidance-actuation latency below 100ms — 42ms at the median and 78ms at the 99th percentile [uploaded benchmark report, p.14] — driving the avoidance controller at an effective 240Hz [src-4] (roughly 10× the usable frame rate of a conventional frame-based vision stack) with field-tuned suppression of fluorescent-flicker event-rate overload [src-5], and fast enough to bring an autonomous mobile robot to a safe stop within the stopping distance required by OSHA 1910.178 [regulatory note, attorney-supplied] for mixed human/robot aisles.
How it works
The event camera produces asynchronous pixel events — tuples of (x, y, polarity, timestamp) — at sub-millisecond latency, with a native rate that scales with scene motion rather than a fixed frame clock. Simultaneously, the solid-state LiDAR emits a conventional depth frame at 10Hz [architecture doc §3.2]. The alignment kernel, running on a dedicated FPGA fabric, maintains a rolling 12ms buffer of events keyed on hardware-clock timestamps [internal whitepaper §2] and projects each event into the nearest LiDAR frame’s voxel grid using a learned epipolar correspondence.
Once aligned, each incoming event carries a depth estimate within ±4cm at 5m range [uploaded benchmark report, p.14]. This augmented event stream is consumed directly by a reactive avoidance controller — a purely reactive module, no map — that evaluates a commanded trajectory against the evolving obstacle field every 4.2ms on average [internal whitepaper §3]. When the event camera saturates (luminance transients, direct sunlight at loading-dock doors), the controller falls back to LiDAR-only at 10Hz with reduced commanded speed, gated by a hysteresis threshold to prevent oscillation [founder intake Q15].
The prevailing trajectory in robotics perception over the past five years has been to increase frame-based camera rates — 120fps, then 240fps global-shutter CMOS [industry survey, founder reference] — to close the latency gap. The present invention instead abandons the frame as the fundamental unit of vision, treating the event stream as primary and the LiDAR frame as a secondary correction signal. A practitioner of ordinary skill would not have been motivated to invert the relationship: event cameras are widely perceived as supplementary sensors, useful for high-speed specialty applications (ball tracking, spark inspection) rather than as the backbone of warehouse perception.
Further, the alignment kernel operates on heterogeneous timebases that were previously considered incompatible at production scale. The closest analogues in the literature ([ref-3], [ref-4]) synchronize event streams against conventional CMOS video, which shares a global shutter; alignment against a rolling-acquired LiDAR depth frame is materially harder, and the ±4cm accuracy we achieve [uploaded benchmark report, p.14] is not anticipated by any single reference.
§ Filing clock
Bar-date signals — attorney determination requiredDates below, including any recommended filing-by date, illustrate one demo company’s disclosure timeline — not a determination for your own matter. Attorney confirmation is required before treating any date or recommendation shown here as applicable to your filing decisions.
The CES demo is the binding §102(b) bar: a public disclosure starts the 12-month clock regardless of filing intent. When to file the provisional ahead of that date — including any internal buffer — is counsel’s determination.
§ Measurable advantages
Quantitative performance claims derived from uploaded evidence. These same quantitative results can serve two independent purposes: as working examples supporting a §112(a) enablement showing, and — as a distinct §103 argument — as objective indicia of non-obviousness (unexpected results / commercial success under Graham v. John Deere, 383 U.S. 1 (1966)), the latter only where a nexus to the claimed invention is established.
240 Hz (~10× the usable frame rate of a conventional frame-based vision stack)
Timestamp alignment threshold
8 ms
Reactive controller evaluation cadence
4.2 ms average
§ Claim-element support matrix
Attorney determination requiredCount of independent supporting sources against this section’s configured evidence floor. A count of the record assembled here — not a sufficiency, corroboration, or patentability assessment. Attorney judgment reserved.
Claim-element support matrix pending — produced on the next pipeline run once evidence has been attached to each claim element.
Pass 2 · Verification
Rubric scores appear here once the full analysis pipeline has run.
§ Novelty snapshot
Five claim elements mapped to the closest prior-art reference we found. Three of five are not anticipated in any single reference — the novelty delta below. Hover any [ref-N] to expand the full citation slot.
References below are pending human verification against USPTO records. Only verified references appear in the final report.
Claim element
Closest prior art
Event-camera sensor providing asynchronous pixel events at <1ms latency
describes event-camera pixel output but not downstream fusion with LiDAR depth.
Solid-state LiDAR depth sensor at 10Hz update
discloses the LiDAR hardware class but at 5Hz update and without event-stream alignment.
Timestamp alignment kernel matching event streams to LiDAR frames within 8ms
does NOT disclose the timestamp alignment kernel between event events and LiDAR frames within 8ms.
Novel
Reactive avoidance controller running at 240Hz effective rate
describes reactive controllers at 60Hz but not at a 240Hz effective rate driven by fused asynchronous streams.
Novel
Failure-mode fallback to LiDAR-only at degraded refresh when event camera saturates
does not contemplate a saturation-triggered fallback path between heterogeneous sensor modalities.
Novel
Novelty delta: three of five claim elements are not anticipated in any single existing reference.
§ Prior art mapping · 5 placeholder refs
References below are pending human verification against USPTO records. Only verified references appear in the final report.
Ref
Assignee
Title
Date
Relevance
Key Passage
[ref-1]Pending
Reference pending human verification
Reference pending human verification
—
—
Reference pending human verification
[ref-2]Pending
Reference pending human verification
Reference pending human verification
—
—
Reference pending human verification
[ref-3]Pending
Reference pending human verification
Reference pending human verification
—
—
Reference pending human verification
[ref-4]Pending
Reference pending human verification
Reference pending human verification
—
—
Reference pending human verification
[ref-5]Pending
Reference pending human verification
Reference pending human verification
—
—
Reference pending human verification
Pass 3 · Drafting
§ Eligibility (§101) risk
Risk level: Low
Claim directs to a hardware-software integrated sensor system; subject matter is statutory. Recentive Analytics does not apply (this is a technical improvement, not applied ML on new data).
Automated §101 eligibility risk signal — not a legal opinion. This analysis is generated by an automated system, may be incomplete or incorrect, and does not resolve patentability. An attorney must independently assess eligibility before any filing or abandonment decision.
§103 obviousness · pre-empt the combinations
Attorney determination required§103 combinations below are signals an examiner could articulate — not a patentability verdict. Attorney weighs each combination against the full prosecution record.
Three combinations of the cited references, taken together, get closer to teaching all claim elements than any single reference. Pre-empting these in the spec saves one office-action round on §103.
Examiner would argue [ref-1] teaches the async event-pixel output and [ref-2] teaches the 10Hz solid-state LiDAR — combining them gives a heterogeneous sensor suite. Silent on the timestamp-alignment kernel (8ms) and saturation-fallback (claim elements 3 + 5).
Applicant counter-argument
No teaching/suggestion/motivation to combine: [ref-1] presents event cameras as standalone high-speed vision, [ref-2] presents LiDAR as standalone depth. The closest analogues ([ref-3], [ref-4]) align events against CMOS video (global-shutter) — alignment against rolling-acquired LiDAR is materially different and the POSA would have been deterred by the heterogeneous timebase, not motivated to solve it.
Secondary indicia (Graham v. Deere)
±4cm depth accuracy at 5m range is unexpected — the literature consensus was that event-to-LiDAR alignment below ±10cm was infeasible at production scale
The sub-100ms stopping distance required by OSHA 1910.178 for mixed aisles was a long-felt need unmet by either reference alone
[ref-3] teaches some form of multi-modal timestamp alignment and [ref-4] teaches a reactive obstacle controller. Examiner could argue combining them yields claim elements 3 + 4 together.
Applicant counter-argument
[ref-3]'s alignment is against CMOS video (global-shutter, single timebase) — NOT the rolling-acquired LiDAR timebase the present claim recites. [ref-4]'s reactive controller runs at 60Hz on frame-based input, not 240Hz on fused asynchronous streams. The specific 8ms threshold and 240Hz effective rate are measurable on a deployed system and are not disclosed by either ref at those bounds.
Secondary indicia (Graham v. Deere)
Commercial skepticism: event cameras were widely perceived in the industry as supplementary sensors for high-speed specialty applications, not as the backbone of warehouse perception
[ref-2] LiDAR + [ref-5] generic fallback schemes
2-ref combination
[ref-2][ref-5]
What the examiner might argue
Combining the LiDAR hardware class with generic fallback-on-sensor-failure teachings arguably reaches claim element 5 (saturation-triggered LiDAR-only fallback).
Applicant counter-argument
[ref-5]'s fallback schemes operate on homogeneous sensor pairs (camera-to-camera, LiDAR-to-LiDAR) and do not contemplate a saturation-triggered fallback between heterogeneous modalities. The hysteresis gate preventing controller oscillation is a specific engineering solution the POSA would not arrive at by mere combination — it's a recognized problem only identified once a heterogeneous fusion stack is deployed.
§ Enablement (§112(a)) detail
Enablement (§112) detail is shown here when the audit has scored the enablement factors for a candidate. This section reflects only what the audit has recorded for your submitted materials.
§ Inventor & rights information
Named inventors with contribution descriptions, employment/assignment status, and funding sources. Inventorship errors are correctable under 35 U.S.C. § 256 — but uncorrected or deceptive errors can jeopardize a patent’s enforceability. Counsel should resolve every yellow or red flag before the provisional is filed.
Arjun Patel
CTO · perception architecture lead
Clean
Contribution
Conceived the inverted-frame approach (event stream as primary, LiDAR as correction); authored the alignment-kernel mathematical framework.
Assignment
Cortex Robotics · IIA signed 2025-08-12
Funding source
Self-funded R&D · no government contract overlap
Mae Johansson
Lead robotics engineer · sensor fusion
Unverified
Contribution
Implemented the timestamp alignment kernel and the epipolar correspondence projection. Hardware-in-the-loop testing.
Assignment
Cortex Robotics · IIA signed 2025-08-12
Funding source
Self-funded R&D
Joined from NVIDIA Aug 2024. Verify their reactive-controller work at NVIDIA does not overlap with Direction B claims; founder interview required.
Devon Reeves
Embedded systems · FPGA implementation
Prior-employer / funding exposure
Contribution
Designed and implemented the FPGA fabric for the rolling event buffer; real-time OS integration.
DARPA funding clause requires government-purpose license review before filing. Counsel must clear FAR 52.227 march-in rights against the claimed FPGA implementation.
Lin Wei
Senior software engineer · reactive controller
Clean
Contribution
Built the reactive avoidance controller and the saturation-fallback hysteresis logic.
Assignment
Cortex Robotics · IIA signed 2024-11-04
Funding source
Self-funded R&D
Inventorship gate: 4 inventors named, 2 require clearance. Counsel sign-off needed on Mae Johansson’s prior-employer scope and Devon Reeves’s DARPA funding clause before the provisional is filed.
§ Filing strategy · 5 dimensions beyond the provisional track
Filing-strategy signals — attorney determination requiredEach dimension below — including any recommended action, filing path, or date in its headline and rationale — describes what this analysis scenario surfaced, not a determination for your matter. Attorney confirmation is required before treating any dimension, urgency label (including “File now”), path, or date shown here as an instruction to act.
Beyond the provisional + utility track, five decisions carry multi-year consequences for the Cortex patent portfolio. Pre-answered here so counsel can ratify vs. negotiate instead of investigate from scratch.
PCT strategy
This year
File a PCT application within 12 months of the provisional — national phase in EP, JP, and optionally KR at 30 months.
Cortex's near-term commercial markets are the US + EU (Amazon/DHL/Maersk warehouses), Japan (Mitsubishi + Daifuku automation), and Korea (Coupang). Direct Paris Convention filings in all three cost ~$18K at 12-month mark; a single PCT buys 18 additional months of national-phase flexibility for ~$5K and lets the Series B read-out inform which markets are worth the per-jurisdiction $6K. Skip Canada + AU national phase — ROI doesn't support for robotics-platform IP.
Paris Convention priority
This year
Foreign filings must lodge within 12 months of Jun 3, 2026 provisional — June 3, 2027 is the hard bar in absolute-novelty jurisdictions.
EP, JP, CN, and KR are absolute-novelty — the CES 2027 demo (Jan 6, 2027) triggers a §102 loss on that date UNLESS a Paris Convention priority claim traces back to the US provisional. As long as the PCT (or direct national filings) cite the provisional within 12 months, the CES disclosure doesn't bar foreign filings. Post-June 2027, foreign rights are lost regardless of US filing status.
Design patents
This quarter
File a design patent on the Cortex AMR's cowling + aisle-facing sensor cluster — separate $2K filing, 15-year term, orthogonal to utility claims.
Cortex's robot form factor is visually distinctive (the angled sensor skirt + LiDAR crown are recognisable in customer videos). Design patents protect the ornamental aspect — a competitor shipping a utility-patent workaround can still be blocked if their unit copies the look. $2K filing, no maintenance fees until year 14, and infringement analysis is visual rather than element-by-element. File alongside the utility provisional so the §102(b) bar clocks run in parallel.
Continuation vs. CIP
Watch
Continuation (not CIP) for mid-PCT-window improvements; separate provisional + utility for post-PCT breakthroughs.
Continuations preserve the earliest priority date for any material also disclosed in the parent — critical against post-June-2027 prior art. Continuation-in-part (CIP) carries a priority-break trap: only the original material keeps the parent's date; new material gets the CIP's later filing date, which can be fatal if a competitor publishes in between. Rule of thumb: if an improvement is supported by the original spec, use a continuation; if it's a genuinely new concept (e.g., a cross-fleet coordination protocol discovered after the provisional), file it as its own track.
Trade-secret trade-off
File now
File cand-001, cand-003, cand-005. Hold cand-002, cand-008 as trade secrets — reasoning below.
cand-001 (sensor fusion kernel) and cand-003 (thermal scheduler) are high-detectability (observable via product tear-down or kernel trace at a customer site) — patents are the right instrument. cand-005 (safety envelope) is a regulatory-defensibility play worth filing. cand-002 (RL warehouse planner) has the opposite profile: near-impossible to detect from the outside (observable only from training data + model weights), and a patent would TELL competitors the approach. Trade-secret protection with NDAs on customer demos beats patents for cand-002. cand-008 is a Recentive-exposed ML application — publish a Nature Machine Intelligence brief for defensive citation, don't file.
§ Missing information & open questions
Pre-written agenda for the founder interview. Walk into the inventor meeting with these and the 60-minute review becomes a 30-minute decision session, not a 3-hour clarification loop.
High · 3Medium · 3Low · 1
High
Confirm DARPA SBIR Phase II contract allows private patent ownership for the FPGA implementation, or whether government-purpose license applies under FAR 52.227-11.
Ask: Devon Reeves · counsel to review SBIR contract language before filing
High
Does Mae Johansson’s reactive-controller work at NVIDIA (Aug 2023 – Aug 2024) overlap with the 240Hz controller we’re claiming in Direction B?
Ask: Mae Johansson · produce employment IIA, NVIDIA project list
High
Is the 8ms timestamp alignment threshold a hard claim limit, or would you accept ≤10ms or ≤15ms to widen claim scope?
Ask: Arjun Patel · founder preference between scope and defensibility
Medium
Provide vendor data sheet for the 1MP event camera (Prophesee Gen 4? Sony IMX636?) and approximate per-unit cost — needed for cost-of-implementation defensibility paragraph.
Ask: Arjun Patel · engineering procurement record
Medium
Confirm the “42ms p50 / 78ms p99” sensor-photon-to-avoidance-actuation figures were measured with full pick-loop in flight, or in standalone perception bench. Different number changes claim language.
Ask: Devon Reeves · benchmark methodology, pp. 12–14
Medium
Any third-party libraries used in the alignment kernel that require attribution, LGPL/GPL exemption, or patent grant disclosures?
Ask: Devon Reeves · SBOM
Low
Should we file a separate continuation on the saturation-fallback hysteresis logic (claim element 5), or roll it into the parent application?
Ask: Counsel decision after Direction B prior-art clearance
§ Disclosure timeline
Bar-date signals — attorney determination requiredEvents below are derived from recorded milestones. “Is bar trigger” marks events that may start the US §102(b) one-year grace clock. EP, JP, and CN jurisdictions give no grace period (absolute novelty). Attorney confirmation required before acting on any bar-date implication.
Provisional deadline
Jun 3, 2026
Internal target — file before the CES public disclosure.
CES 2027 demo
§102(b) clock eventJan 6, 2027
Public disclosure — starts the §102(b) 12-month clock.
Utility deadline
Jun 3, 2027
12 months after the provisional — Paris Convention foreign-filing bar.
§ Foreign-filing exposure signals
Attorney determination requiredExposure signals only — not jurisdiction recommendations. The “absolute novelty exposure” dimension reflects doctrine (EP/JP/CN absolute-novelty bar); the remaining four are commercial-signal inputs for attorney-driven jurisdiction strategy. No filing act (“file in EP”, “file PCT by [date]”) is expressed or implied. CA rows carry CA §28.2 one-year grace from the Canadian filing date.
Absolute novelty exposuredoctrine
The CES 2027 demo (Jan 6, 2027) triggers a §102 novelty loss in absolute-novelty jurisdictions unless a Paris Convention priority claim traces back to the US provisional.
Market presence signal
US + EU (Amazon / DHL / Maersk warehouses), Japan (Mitsubishi + Daifuku automation), Korea (Coupang).
Manufacturing exposure
Self-funded R&D; no disclosed foreign manufacturing footprint yet.
§ Trade-secret note
Trade-secret vs. patent election — attorney determination requiredThis candidate was system-classified as not recommended for patent filing. Whether to protect it as a trade secret instead of a patent is a legal election for your patent attorney to make — this platform does not advise you to pursue, or to forgo, trade-secret protection. The note below describes the system's classification and how the export was rendered; it is not legal advice and not a recommendation to elect trade-secret protection.
Two candidates are better protected as trade secrets than as patents. cand-002 (RL warehouse planner) is near-impossible to detect externally — observable only from training data and model weights — and a patent would disclose the approach to competitors; hold it as a trade secret with NDA-gated customer demos. cand-008 is a Recentive-exposed ML application — publish a defensive brief for citation rather than filing. File cand-001, cand-003, and cand-005.
Informational analysis onlyThis is an automated, informational analysis of the materials you submitted. It is not legal advice and does not create an attorney-client relationship. The candidate concepts identified are signals for attorney review — not patentability determinations. Consult a registered patent attorney or agent before making any filing or IP-strategy decisions.
AI inventorship disclosure
This PRP was generated with AI assistance. Named inventors above are human contributors; AI tools contributed analytical support, not inventive conception. Full AI model audit trail (prompts, model version, retrieval corpus) is available on request — it supports the inventorship analysis and any duty-of-disclosure obligations (37 CFR 1.56), and is not itself a USPTO-mandated filing.