Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 05/11/2026 has been entered.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
Applicant's arguments and remarks filed on 04/13/2026 have been fully considered.
Applicant's Request for Continued Examination (RCE) filed on 05/11/2026 have been fully considered.
Applicant's amendments overcome the objections to the claims.
Claims 1, 2, 8 and 15 have been amended.
No new matter has been introduced.
Claims 4-6, 11-13 and 18-20 are objected to.
Claims 1-20 are pending.
Response to Arguments
Applicant’s arguments with respect to claims 1-3, 7-10, 14, and 15-17 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-3, 7-10, and 14-17 are rejected under 35 U.S.C. 103 as being unpatentable over Sun et al. (Real-Time Radar-Based Gesture Detection and Recognition Built in an Edge-Computing Platform. IEEE Sensors Journal, 20(18), 10706-10716. https://doi.org/10.1109/JSEN.2020.2994292) ("Sun et al.") in view of Wang et al. (A Novel Detection and Recognition Method for Continuous Hand Gesture Using FMCW Radar, IEEE Access, vol. 8 (2020) (“Wang et al.”).
Regarding Claims 1, 8, and 15, Claims 1, 8, and 15 are independent claims directed to a method, an electronic device, and a non-transitory computer readable medium, respectively. The bodies of Claims 8 and 15 recite substantively identical functional limitations as Claim 1, differing only in their preambles and statutory class. Claims 1, 8, and 15 are therefore grouped and the full analysis is presented for Claim 1. Sun et al. in view of Lien et al. (‘853) teaches:
Sun et al. teaches: A method implemented by at least one processor, the method comprising: obtaining a stream of radar data into a sliding input data window that is composed of radar frames from the stream, each radar frame within the data window including features selected from a predefined feature set and at least one of time-velocity data (TVD) or time angle data (TAD) (Pg. 10707: “For every measurement-cycle, we store this information in a feature matrix with reduced dimensions”; Pgs. 10708-10709, Eq. (8): “we encode the range, Doppler, azimuth, elevation and magnitude of those K points with the largest magnitudes in RD(p, q) along IL measurement-cycles into the feature cube V”). Sun et al.’s per-measurement-cycle feature matrices each store a selected set of features (range, Doppler, azimuth, elevation, magnitude) and constitute radar frames; the range/Doppler features constitute time-velocity data (TVD) and the azimuth/elevation features constitute time angle data (TAD). The stream of measurement-cycles encoded along IL cycles into the feature cube V constitutes the stream of radar data obtained into a sliding input data window.
Sun et al. teaches: for each radar frame within the data window, receiving a binary prediction indicating whether the radar frame includes a gesture end (pg. 10709: “We propose a STA/LTA-based gesture detector to detect the tail of a gesture”; pg. 10709: “our proposed HAD algorithm is designed to detect when a gesture finishes, i.e., the tail of a gesture, rather than detecting the start time-stamp”). At each measurement-cycle the STA/LTA detector evaluates whether the tail of a gesture (the gesture end) is detected, producing a two-valued outcome per frame that constitutes the binary prediction indicating whether the radar frame includes a gesture end.
Sun et al. teaches: in response to the binary prediction indicating that the radar frame includes the gesture end, triggering an early stop checker to determine whether an early stop condition is satisfied by the data window, wherein determining whether the early stop condition is satisfied comprises determining whether a noise frames condition and a valid activity condition are satisfied (pg. 10709, Eq. (12), requiring both conditions to be fulfilled to detect the tail; pg. 10710: “one condition of detecting the tail of a gesture is that, the average of RWM in the long window exceeds the threshold γ1. It means that a hand motion appears in the long window”; pg. 10710: “The other condition is that, the ratio of the mean EMA in the short window and that in the long window is lower than the threshold γ2 … it detects when the hand movement finishes”). Upon detecting the tail of a gesture, Sun et al. evaluates two conditions that together determine whether to forward the gesture data, which constitutes triggering an early stop checker to determine whether an early stop condition is satisfied. The condition that hand movement has finished (γ2) constitutes the noise frames condition, and the condition that a hand motion appears in the long window (γ1) constitutes the valid activity condition.
Sun et al. teaches: in response to a determination that the early stop condition is satisfied, triggering a gesture classifier (GC) to predict a gesture type (Pg. 10707: “After detecting the tail of a gesture, we arrange the feature matrices belonging to the measurement-cycles, which are previous to this tail, into a feature cube … It is subsequently fed into a shallow CNN for classification”; pg. 10709, Fig. 4(b): “The gesture data is directly forwarded to the classifier without delay when we detect the tail of the gesture”). Upon satisfaction of the dual condition of Eq. (12), Sun et al. forwards the feature cube to the CNN classifier, which constitutes triggering a gesture classifier to predict a gesture type.
Sun et al. teaches: wherein the determination that the early stop condition is satisfied includes a determination that the noise frames condition is satisfied based on the data window including a lookback window of w radar frames that are noise frames, where w is a non-zero positive integer (pg. 10709, Eq. (11), defining STA(l) as the mean EMA over the short window of length L1; pg. 10710, Eq. (12); pg. 10710: “it detects when the hand movement finishes”). Sun et al.’s short window of length L1 constitutes the lookback window of w radar frames, where w = L1 is a non-zero positive integer; the second condition (the short-window-to-long-window ratio being below γ2) being satisfied indicates the most recent L1 frames are quiescent frames in which hand movement has finished, which constitutes the determination that the lookback window of w radar frames are noise frames.
It would have been obvious to a person having ordinary skill in the art (PHOSITA) before the effective filing date of the claimed invention to express Sun et al.’s per-measurement-cycle feature matrices as the claimed “radar frames” obtained into a “sliding input data window.” Sun et al. stores hand information “for every measurement-cycle” (pg. 10707) and processes a running succession of such measurement-cycles, which is the same frame-indexed, windowed processing recited in the claim. One may argue that the claimed “radar frame / sliding input data window” vocabulary is not explicitly used by Sun et al., however, Wang et al. teaches representing FMCW radar gesture data as per-frame spectrogram values processed over a frame window (Wang et al., p. 167268: “we propose a simple but effective continuous hand gesture detection method based on the amplitude of hand gesture spectrograms”). One would have been motivated to combine Sun et al. and Wang et al. because both are directed to FMCW radar hand-gesture recognition using range, Doppler, and angle features over frame windows, and Wang et al.’s frame-indexed representation predictably supplies the explicit per-frame windowing convention. There is a reasonable expectation of success because the two references operate on the same kind of radar data in the same field. The frame-indexed windowed processing is expressly present in both Sun et al. and Wang et al.
Claims 8 and 15 are rejected for the same reasons as Claim 1, as their claim bodies are substantively identical to Claim 1; the “processor” of Claim 8 and the “program code that when executed causes at least one processor to” of Claim 15 are taught by Sun et al.’s implementation on an edge-computing platform (pg. 10710: “The proposed multi-feature encoder and HAD have been implemented … while the implementation of the CNN is supported by TensorRT”).
Regarding Claims 2, 9, and 16, the claims 2, 9, and 16 depend from Claims 1, 8, and 15, respectively, and each recite the same additional limitations. They are grouped and the analysis for Claim 2 is presented below; Claims 9 and 16 are rejected for the same reasons, with their respective parent claims providing the foundational rejection. Regarding Claim 2, Sun et al. in view of Wang et al. teaches the method according to Claim 1.
Sun et al. teaches: for each radar frame within the data window, receiving an accumulator status (c) that is mapped to a lookback window size w and to a feature threshold (fth) that corresponds to each of the selected features (pg. 10709: “L1 and L2 are the length of the short and long window”; pg. 10710: “the parameters β, γ1 and γ2 in our HAD algorithm should be thoroughly chosen according to different application scenarios”). Sun et al.’s short-window length L1 constitutes the lookback window size w, and the detection thresholds γ1 and γ2 applied to the windowed feature values constitute the feature thresholds fth corresponding to the selected features.
Sun et al. teaches: determining that the early stop condition is satisfied based on a determination that both the noise frames condition and the valid activity condition are satisfied (pg. 10709-10710, Eq. (12): “The tail of a gesture is detected, when the following conditions are fulfilled”, the two conditions being joined by “and”). The tail of a gesture is detected only when both Sun et al.’s conditions are fulfilled together, which constitutes determining the early stop condition is satisfied based on both the noise frames condition and the valid activity condition being satisfied.
Sun et al. teaches: determining the noise frames condition is satisfied when the lookback window of w radar frames in the data window are noise frames in which the selected features are less than the corresponding fth (pg. 10710: “it detects when the hand movement finishes”, referring to the condition STA(l)/LTA(l) ≤ γ2 of Eq. (12)). The short-window-to-long-window ratio falling below γ2 indicates that the magnitude in the recent L1 (i.e., w) frames has dropped below the level set by the threshold, which constitutes determining the recent w frames are noise frames in which the selected features are less than the corresponding threshold.
Sun et al. teaches that hand-motion activity is present in the data window: (pg. 10710: “the average of RWM in the long window exceeds the threshold γ1. It means that a hand motion appears in the long window”). Sun et al.’s first condition establishes that a valid hand motion is present within the data window.
Sun et al. does not explicitly teach, but Wang et al. teaches: determining the valid activity condition is satisfied when the data window contains a valid activity within at least a data window threshold number of radar frames that are outside the lookback window and are not noise frames (Wang et al., p. 167268: “when the amplitude is first larger than the threshold, we mark it as the starting frame of the hand gesture. The hand gesture is ended when the Ysumi is no larger than the threshold”; Wang et al., p. 167268, Eq. (26): “Garea = Gesture frame Ysumi > threshold”). Sun et al.’s first condition (γ1) tests only whether an average over the long window exceeds a level, and Sun et al.’s short and long windows are nested rather than disjoint; Sun et al. therefore does not determine, on a per-frame basis, which individual frames are signal frames, nor require that valid-activity signal frames be located outside the lookback window of noise frames. Wang et al. supplies this: Wang et al. determines for each frame whether the frame is a gesture (signal) frame by comparing the frame’s summation amplitude to a threshold, marking a frame as a gesture frame when its amplitude exceeds the threshold and treating frames whose amplitude is not larger than the threshold as outside the gesture, and identifies the gesture as a bounded run of consecutive above-threshold signal frames terminated by the below-threshold (noise) frames — for example, “the duration time of the first hand gesture is 16 frames (from the 10-th frame to the 25-th frame)” (Wang et al., p. 167268). That bounded run of above-threshold signal frames constitutes radar frames that are not noise frames and that are located outside the trailing below-threshold frames marking the gesture end, corresponding to the claimed valid activity within radar frames that are outside the lookback window and are not noise frames.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to combine Sun et al.’s STA/LTA tail-detection scheme with Wang et al.’s per-frame amplitude-threshold gesture segmentation. One would have been motivated to do so because Wang et al. teaches, in the same FMCW radar gesture-recognition context as Sun et al., that comparing each frame’s amplitude to a threshold to mark signal frames and bounding the gesture as a run of such frames effectively segments continuous hand gestures and rejects non-gesture frames, reporting that “the accuracy of the proposed gesture detection method can reach 96.17%” (Wang et al., Abstract). Incorporating Wang et al.’s per-frame signal-frame determination into Sun et al.’s detector would predictably ensure that genuine signal frames distinct from the end-of-gesture quiescent frames are present before the classifier is triggered, reducing false triggers. There is a reasonable expectation of success because Sun et al. and Wang et al. both operate on FMCW radar gesture data represented as range, Doppler, and angle features over frame windows, so Wang et al.’s per-frame thresholding integrates directly with Sun et al.’s windowed detection. The rationale does not rely on impermissible hindsight because the per-frame signal/noise threshold determination is expressly taught by Wang et al.
Claims 9 and 16 are rejected for the same reasons as Claim 2, as their claim bodies are substantively identical to Claim 2.
Regarding Claims 3, 10, and 17, the claims 3, 10, and 17 depend from Claims 1, 8, and 15, respectively, and each recite the same additional limitations. They are grouped and the analysis for Claim 3 is presented below; Claims 10 and 17 are rejected for the same reasons, with their respective parent claims providing the foundational rejection.
Sun et al. teaches: determining whether to output an event indicator indicating that a user of an electronic device performed the gesture type predicted, based on whether a stop confirmation condition is satisfied by a GC output that includes the gesture type predicted; and in response to determining the stop confirmation condition is satisfied, outputting an indicator that the GC output satisfied the stop confirmation condition (pg. 10707: “It is subsequently fed into a shallow CNN for classification”; pg. 10710, identifying the 12 output gesture classes (a) Check through (l) Swipe right, the CNN output layer using “a softmax function” that “normalizes the output of the last FC layer to a probability distribution over the classes”). Sun et al.’s classifier produces, for a detected gesture, an output identifying the predicted gesture class, and the system reports that recognized gesture; conditioning the report on the classifier producing a valid recognized-gesture output constitutes determining whether to output an event indicator based on a stop confirmation condition satisfied by the GC output, and reporting the recognized gesture constitutes outputting the indicator.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to output an indicator of the recognized gesture, conditioned on the classifier producing a valid gesture-class output, in the Sun et al./Wang et al. system. One would have been motivated to do so because the purpose of Sun et al.’s system is to recognize and report gestures in real time for device interaction (pg. 10707), and outputting the recognized gesture only when the classifier yields a valid class predictably avoids reporting spurious results. There is a reasonable expectation of success because Sun et al.’s softmax classifier already produces a per-class output suitable for such a determination. The rationale does not rely on impermissible hindsight because the gesture-class output and its reporting are expressly taught by Sun et al.
Regarding Claims 7 and 14, the claims 7 and 14 depend from Claims 1 and 8, respectively, and each recite the same additional limitations. They are grouped and the analysis for Claim 7 is presented below; Claim 14 is rejected for the same reasons.
Regarding Claim 7, Sun et al. in view of Lien et al. (‘853) teaches the method according to Claim 1.
Sun et al. teaches: for each radar frame within the data window, receiving an accumulator status (c); and skipping triggering the GC based on a determination that the c is not equal to a multiple of a multiplier (k) for controlling periodicity of triggering the GC (pgs. 10708-10709: “the classifier is only activated when a gesture is detected rather than keeping it active for every measurement-cycle”). Sun et al. activates the classifier only upon a detection event rather than at every measurement-cycle, thereby controlling the periodicity of classifier triggering and skipping classifier triggering for frames that do not satisfy the detection condition.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to control the periodicity of classifier triggering by a count-based condition (triggering only when an accumulated frame count equals a multiple of a multiplier k, and skipping otherwise) in the Sun et al./Wang et al. system. One would have been motivated to do so for Sun et al.’s express reason of designing “a power-efficient gesture recognition system” (pg. 10709) by limiting how often the classifier is activated; implementing that gating as a count-multiple condition is a predictable design choice yielding fewer classifier invocations. There is a reasonable expectation of success because Sun et al. already conditions classifier activation on a detection event in the same system. The rationale does not rely on impermissible hindsight because limiting classifier-activation frequency for power efficiency is expressly taught by Sun et al.
Allowable Subject Matter
Claims 4 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
The following is a statement of reasons for the indication of allowable subject matter. The closest prior art of record - Sun et al. (Real-Time Radar-Based Gesture Detection and Recognition Built in an Edge-Computing Platform. IEEE Sensors Journal, 20(18), 10706-10716. https://doi.org/10.1109/JSEN.2020.2994292 (Year: 2020)) (STA/LTA tail detection with a two-condition gate triggering a CNN classifier), Hong et al. (US 2022/0261084 A1) (robust gating and false-positive suppression), Stern et al. (US 2021/0064146 A1) (paused-gesture handling), Jeppsson et al. (US 2021/0103348 A1) (gesture proficiency/feedback), Wang et al. (A Novel Detection and Recognition Method for Continuous Hand Gesture Using FMCW Radar. IEEE Access, 8, 167264–167275. https://doi.org/10.1109/ACCESS.2020.3023187) (FMCW RTM/DTM/ATM segmentation by decision threshold), and Lien et al. (US 2016/0320853 A1)/Poupyrev et al. (US 2020/0064924 A1) (radar-frame velocity processing) - teaches detecting the end of a gesture, gating a classifier on activity/quiescence conditions, and reporting a recognized gesture. None of these references, alone or in combination, teaches or suggests the following:
Claims 4, 11, and 18. These claims require that, upon the stop confirmation condition not being satisfied, the early stop checker is updated by reconfiguring the noise frames condition and the valid activity condition to be gesture-based on the tentatively predicted gesture type, and that the system thereafter selects between gesture-based and general versions of those conditions depending on whether the early stop checker has been updated, re-triggering the classifier under the gesture-based conditions. The art of record applies fixed, gesture-agnostic detection conditions. Sun et al. expressly contemplates only that its parameters be chosen in advance “according to different application scenarios”; it does not feed a tentative classifier output back to redefine the detection conditions for the same gesture in progress, nor maintain dual (general versus gesture-based) condition sets selected at runtime. This closed-loop, classifier-informed updating of the early stop conditions is neither disclosed nor suggested.
Claims 5, 12, and 19. These claims require maintaining a subset of pause-free gestures (G2S) and, when the predicted gesture type is within that subset and the GC output includes a derived gesture length (lg) falling within a specified range corresponding to that gesture type, outputting the event indicator without evaluating the stop confirmation condition. The art recognizes that gesture durations differ and that confirmation reduces false positives, but none of it conditions a confirmation bypass on joint satisfaction of (i) membership in a pause-free gesture subset and (ii) a per-gesture-type gesture-length range. The specific gesture-length-gated shortcut keyed to a defined pause-free subset is not taught or suggested.
Claims 6, 13, and 20. These claims require maintaining a subset of gestures (G2M) that include a pause or repeat-motion and, based on whether the predicted gesture type is within that subset, either outputting the event indicator immediately (when outside G2M) or triggering a waiting window and outputting only after it elapses (when within G2M). While Stern teaches a general paused-gesture state, it does not classify gesture types into a pause/repeat-motion subset and trigger a type-conditioned waiting window before reporting; the remaining references do not cure this. Sun et al. in fact identifies the pause/turning-point problem (e.g., “Cross”) as a source of error without proposing a subset-keyed waiting-window remedy. The waiting-window mechanism selectively applied to a defined pause/repeat-motion gesture subset is not taught or suggested.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to REMASH R GUYAH whose telephone number is (571)270-0115. The examiner can normally be reached M-F 7:30-4:30.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Resha H Desai can be reached at (571) 270-7792. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/REMASH R GUYAH/Examiner, Art Unit 3648
Prosecution Insights