DETAILED ACTION
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 .
Priority
Acknowledgment is made of applicant's claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. EP23175096.9, filed on 05/24/2023.
Response to Amendment
Applicants' arguments and remarks filed on 06/18/2026 have been fully considered.
Claims 1, 4-8, 11-16, and 18 have been amended.
Claims 3, 10, and 19 have been canceled.
Claims 20-23 are new.
Claims 1, 2, 4-9, 11-18, and 20-23 are pending
Response to Arguments
Applicant’s arguments, see remarks pages 10-13, filed 06/18/2026, with respect to Claims 1-19 have been fully considered and are persuasive. The 35 U.S.C. 103 rejection of Claims 1-19 has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of:
Claims 1, 5, 6, 9, 13-16, 18, and 20-22: Weisenburger et al. (US 2022/0018973 A1) in view of Werner et al. (US 2020/0049837 A1).
Claims 4 and 11: Weisenburger et al. (‘973) in view of Werner et al. (‘837), and further in view of Yue Zhe et al. (CN 116088012 A).
Claim 2: Weisenburger et al. (‘973) in view of Werner et al. (‘837), and further in view of Kim et al. (US 2020/0132861 A1).
Claim 12: Weisenburger et al. (‘973) in view of Werner et al. (‘837) and Yue Zhe et al. (’012), and further in view of Kim et al. (US 2020/0132861 A1).
Claim 17: Weisenburger et al. (‘973) in view of Werner et al. (‘837), and further in view of Siercks et al. (US 2020/0240784 A1).
Claim Objections
Claim 1 objected to because of the following informalities:
Claim 18 is objected to as follows. Claim 18 has been amended to recite computer-executable instructions for causing a computer to perform the method according to claim 1. Claim 1 requires receiving, via a GNSS antenna, a plurality of GNSS signals and capturing a digital image using an imaging device, neither of which a computer alone can perform. The amendment therefore does not supply the clarification identified in the prior Office Action. Applicant may overcome the objection by reciting instructions which, when executed by a system comprising a GNSS antenna and an imaging device, cause the system to perform the method according to claim 1.
Claim 22 depends from Claim 20 and recites the potential GNSS signal quality values, determining the potential GNSS signal quality value, and the pixel or the set of coherent pixels corresponding to the respective GNSS satellite. These limitations in Claim 22 or in Claim 20, each is first introduced in Claim 21. For purposes of examination, Claim 22 is interpreted as depending from claim 21. The dependency of Claim 22 on Claim 20 appears to be a clerical error.
Appropriate correction is required.
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, 5, 6, 9, 13-16, 18, and 20-22 are rejected under 35 U.S.C. 103 as being unpatentable over Weisenburger et al. (US 2022/0018973 A1) in view of Werner et al. (US 2020/0049837 A1).
Regarding Claim 1, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches:
Weisenburger et al. (‘973) teaches: A computer-implemented method for processing satellite signals to derive a geospatial position, the method comprising: ([0072]: “FIG. 11 illustrates a method 1100 for processing satellite signals for computing a geospatial position”; [0073]: “Method 1100 may be implemented as a computer-readable medium or computer program product comprising instructions which, when the program is executed by one or more computers, cause the one or more computers to carry out the steps of method 1100”). The recitation particularly fully automatically and in real time is optional language under MPEP § 2111.04 and is not treated as limiting.
Weisenburger et al. (‘973) teaches: receiving, via a GNSS antenna, a plurality of GNSS signals from a plurality of GNSS satellites ([0074]: “At step 1102, a plurality of GNSS signals (e.g., wireless signals 104, 904) are received from a plurality of GNSS satellites”; [0075]: “The plurality of GNSS signals are received via a GNSS antenna (e.g., GNSS antennas 116, 216, 316, 416, 516, 916)”).
Weisenburger et al. (‘973) teaches: capturing a digital image using an imaging device at least partially oriented toward the plurality of GNSS satellites, the digital image comprising a multitude of pixels, a first subset of pixels imaging the sky, and a second subset of pixels imaging obstructions that are at least partly impermissible to GNSS signals ([0076]: “At step 1104, an image (e.g., images 113, 213, 313, 613) is captured using an imaging device (e.g., imaging devices 112, 212, 312, 412). The imaging device may be mounted to the antenna structure. The imaging device may be at least partially orientated toward the plurality of GNSS satellites”; [0049]: “Image 613 is upward facing and shows the vertical hemisphere in a setting with buildings, trees, and open sky”; [0078]: “the image may be segmented such that each pixel of the image is included in one of the plurality of regions”; [0050]: “Region 662-1 corresponds to clouds or open sky and includes objects that provide little RF obstruction… Region 662-3 corresponds to buildings and includes objects that provide significant RF obstruction”). This element presents an “and/or” statement; the alternative relied upon is obstructions that are at least partly impermissible to GNSS signals, taught by Fig. 6B, region 662-3, of Weisenburger et al. (‘973). The reflective-surface alternative need not be addressed.
Weisenburger et al. (‘973) teaches: determining an orientation of the image ([0079]: “At step 1108, an orientation (e.g., orientations 556, 756) of the image is determined”).
Weisenburger et al. (‘973) teaches: computing a geospatial position ([0086]: “At step 1118, the geospatial position is computed based on the processed plurality of GNSS signals”).
Weisenburger et al. (‘973) teaches: the method further comprising, for each of at least a subset of the plurality of GNSS satellites: ([0082]: “At step 1112, the plurality of GNSS satellites are projected onto the image based on the orientation of the image such that a corresponding region is identified for each of the plurality of GNSS satellites”). Weisenburger performs the subsequent per-satellite operations for each satellite of the plurality, which encompasses at least a subset thereof.
Weisenburger et al. (‘973) teaches: extracting signal features from the respective GNSS signal ([0070]: “Digital samples 1034 generated by RF front end 1030 may be sent to a correlator 1042, which may perform one or more correlations on digital samples 1034 using local codes”; [0085]: “the pseudorange computed for a particular GNSS signal is weighted in accordance with the weight”). Weisenburger correlates the received signal to obtain a pseudorange for each individual GNSS signal, which constitutes a signal feature extracted from the respective GNSS signal.
Weisenburger et al. (‘973) teaches: processing the image to extract image features, wherein processing the image comprises applying machine learning to compute an image feature vector or an image embedding vector for at least a third subset of pixels of the image, the third subset of pixels being defined based on a position of the respective GNSS satellite ([0031]: “the captured image can be segmented into multiple regions based on the predicted RF characteristics of objects in the image, and the orbital positions of the GNSS satellites can be projected onto one of the regions… The segmentation can be performed using a machine-learning model, such as a neural network, or using any one of various image segmentation techniques”; [0032]: “The pixel location onto which the orbital position is projected is recorded. Next, the pixel location is compared to the segmented regions to determine in which region the pixel location is located. The identified region is associated with the GNSS satellite and the corresponding GNSS signal”). This element presents an “or” statement; the alternative relied upon is an image feature vector, and the image embedding vector alternative need not be addressed. Weisenburger applies a machine-learning model to the image to compute, for each pixel, a value characterizing the predicted RF-obstruction property of the imaged object at that pixel. The third subset of pixels is the pixel location determined for the respective GNSS satellite by projecting that satellite’s orbital position onto the image, and the image feature computed for that pixel subset is the identified region, which is then associated with that satellite and its signal. The subset is thus defined based on the position of the respective GNSS satellite.
Weisenburger et al. (‘973) teaches: combining the extracted signal features and the extracted image features to receive a feature combination ([0085]: “At step 1116, each of the plurality of GNSS signals is weighted in accordance with a weight for the corresponding region for the corresponding GNSS satellite… In one example, the pseudorange computed for a particular GNSS signal is weighted in accordance with the weight”). The pseudorange extracted from the respective signal is combined with the region identified for that satellite from the image to produce a weighted pseudorange for that satellite. Claim 4 separately requires the feature combination to be a combined feature vector, confirming that a feature combination in claim 1 is not limited to a vector-valued combination.
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: deriving, based on feature combinations of a plurality of GNSS satellites and by applying machine learning, a signal classification and/or an estimated local error of the respective GNSS signal ([0067]: “the machine learning model 412 may also receive the GNSS receiver data 502, including the above-mentioned parameters”; [0068]: “The machine learning model 412 may be configured to output an amount of error for the GNSS receiver data 502 with respect with the GNSS satellite”). This element presents an “and/or” statement; the alternative relied upon is an estimated local error of the respective GNSS signal, and the signal classification alternative need not be addressed. Weisenburger et al. (‘973) determines the extent to which each signal is used by a fixed weight table associated with the identified region and does not apply machine learning to derive a per-satellite error from the per-satellite combination.
Werner et al. (‘837) teaches: wherein computing the geospatial position is based at least on a subset of the GNSS signals and on the signal classification and/or the estimated local error of each GNSS signal of the subset ([0069]: “with respect to a particular GNSS satellite, the machine learning model 412 may determine location estimation error of 100 meters (e.g., high estimation error). Thus, the location estimator 500 may disregard and/or limit measurements corresponding to that GNSS satellite”).
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 replace the fixed, region-indexed weight table by which Weisenburger et al. (‘973) processes each GNSS signal with the trained machine learning model of Werner et al. (‘837), so that the per-satellite combination of pseudorange and identified image region produced by Weisenburger is supplied to the model of Werner as input and the model outputs an estimated error for that satellite that governs how the signal is used in computing the geospatial position. One would have been motivated to do so because both references address the identical problem of degraded pseudorange accuracy caused by obstruction, reflection and diffraction of GNSS signals in urban and vegetated environments (Weisenburger et al. (‘973), [0030]: “Tracking of a blocked satellite indicates that the pseudorange and carrier phase measurements correspond to reflected signals and should either be discarded or deweighted relative to the LOS signals”; Werner et al. (‘837), [0014]: “Challenging signal environments (e.g., urban canyons, areas of dense foliage, areas near or within structures such as buildings, and/or other areas that may interfere with line of sight reception of signals) can complicate the computation of an accurate position solution”), and because Werner expressly identifies the deficiency of the fixed uncertainty values that Weisenburger’s weight table embodies and teaches that a trained model yields a more accurate per-satellite error figure (Werner et al. (‘837), [0068]: “This indication of error (e.g., as output by the machine learning model 412) may be a more accurate indication of error and/or uncertainty than the uncertainty parameters (e.g., pseudorange uncertainty, range rate uncertainty) included with the GNSS receiver data 502”). A PHOSITA reading Weisenburger’s teaching that pseudoranges in the intermediate obstruction region should be deweighted by a coarse tabulated factor would have looked to Werner for a quantitatively graded replacement for that factor. There is a reasonable expectation of success because Werner’s model operates on the same per-satellite quantities that Weisenburger already computes, including pseudorange, elevation above horizon and azimuth (Werner et al. (‘837), [0064]: “Such parameters may include, but are not limited to: pseudorange; pseudorange uncertainty; range rage; range rate uncertainty; a multipath indicator; elevation above horizon; azimuth”), and because Werner already contemplates supplementing those parameters with additional device sensor inputs within the same receiver architecture, so no change to Weisenburger’s antenna structure, imaging device or projection geometry is required.
Regarding Claim 5, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) teaches: projecting, based on the orientation of the image and on known satellite positions, at least a subset of the plurality of GNSS satellites onto the image, so that each projected GNSS satellite corresponds to a pixel or a set of coherent pixels of the digital image ([0082]: “the plurality of GNSS satellites are projected onto the image based on the orientation of the image such that a corresponding region is identified for each of the plurality of GNSS satellites. In some embodiments, a pixel location within the image is first determined for each of the plurality of GNSS satellites”; [0032]: “the GNSS receiver can look up the orbital position of the GNSS satellite that transmitted the signal and project the orbital position onto the captured image”).
Weisenburger et al. (‘973) teaches: determining, for at least a plurality of pixels of the image, a potential GNSS signal quality value ([0077]: “Each region of the plurality of regions may be characterized by an extent to which the plurality of GNSS signals are obstructed by the objects within the region”; [0078]: “the image may be segmented such that each pixel of the image is included in one of the plurality of regions”). The region value assigned to each pixel expresses the extent of obstruction a signal arriving from that direction would experience, which is a potential GNSS signal quality value.
Weisenburger et al. (‘973) teaches: assigning to each of the projected GNSS satellites the potential GNSS signal quality value of the corresponding pixel or the corresponding set of coherent pixels ([0032]: “the pixel location is compared to the segmented regions to determine in which region the pixel location is located. The identified region is associated with the GNSS satellite and the corresponding GNSS signal”).
Regarding Claim 6, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claims 1 and 5, and further teaches:
This claim presents an “and/or” statement joining the limitations below with the reflective-surface limitation; the alternative relied upon is the first, and the reflective-surface alternative need not be addressed.
Weisenburger et al. (‘973) teaches: processing the image comprises applying machine learning to compute an image feature vector for at least the third subset of pixels of the image; and the third subset of pixels is defined based on the position of the respective projected GNSS satellite in the image ([0031]: “The segmentation can be performed using a machine-learning model, such as a neural network”; [0032]: “The pixel location onto which the orbital position is projected is recorded. Next, the pixel location is compared to the segmented regions to determine in which region the pixel location is located”). The pixel location at which the satellite’s orbital position is projected is the third subset, and the machine-learning-computed region value at that location is the image feature for that satellite.
Regarding Claim 9, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 5, and further teaches:
Weisenburger et al. (‘973) teaches: wherein deriving the signal classification and/or the estimated local error is also based on the potential GNSS signal quality values ([0084]: “each of the plurality of GNSS signals is processed in accordance with the corresponding region for a corresponding GNSS satellite of the plurality of GNSS satellites. For example, each of the plurality of GNSS signals may be processed differently based on their corresponding region”). The alternative relied upon is the estimated local error, consistent with claim 1. The per-satellite quality value obtained from the image governs how the corresponding signal is treated, and in the combined system set forth under claim 1 that value is carried into the machine-learning derivation as part of the per-satellite feature combination.
Regarding Claim 13, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) teaches: a signal classification is derived for the GNSS signal of each GNSS satellite of the subset of GNSS satellites; and the geospatial position is computed based on at least a subset of the GNSS signals for which the signal classification is derived and on their respective signal classification ([0082]: “a corresponding region is identified for each of the plurality of GNSS satellites”; [0086]: “the geospatial position is computed based on the processed plurality of GNSS signals”). The optional recitation particularly wherein computing the geospatial position comprises weighting the GNSS signals based on their respective signal classification is not limiting under MPEP § 2111.04 but is noted as taught for completeness ([0085]: “each of the plurality of GNSS signals is weighted in accordance with a weight for the corresponding region for the corresponding GNSS satellite”).
Regarding Claim 14, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 13, and further teaches:
Weisenburger et al. (‘973) teaches: wherein computing the signal classification comprises detecting multipath signals, non-line-of-sight signals and/or diffraction signals ([0030]: “using the GNSS satellite location on the image, it can be determined whether there is a direct visible LOS to the satellite or if it is blocked by an object”). This element presents an “and/or” statement; the alternative relied upon is non-line-of-sight signals, and the remaining alternatives need not be addressed.
This claim then presents an “or” statement between downweighting and exclusion; the alternative relied upon is the second.
Weisenburger et al. (‘973) teaches: the subset of the GNSS signals from which the geospatial position is computed does not comprise any of the detected multipath signals, non-line-of-sight signals and/or diffraction signals, respectively ([0084]: “any GNSS signal that is transmitted from a GNSS satellite that is projected onto the image within a third region is ignored and not used to compute the geospatial position”).
Regarding Claim 15, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: an estimated local error is derived for the GNSS signal of each GNSS satellite of the subset of GNSS satellites; and the geospatial position is computed based on the subset of the GNSS signals for which the estimated local error is derived and on their respective estimated local error ([0069]: “in a case the machine learning model 412 determines a location estimation error of 5 meters (e.g., low estimation error) for a GNSS satellite, the location estimator 500 may prioritize the measurements corresponding to the GNSS receiver data 502 of that GNSS satellite, in determining the estimated device location 506”). The obviousness rationale, motivation to combine and reasonable expectation of success set forth under claim 1 for the combination of Weisenburger et al. (‘973) with Werner et al. (‘837) apply equally here and are expressly incorporated.
Weisenburger et al. (‘973) teaches: wherein the method comprises projecting, based on the orientation of the image and on known satellite positions, at least a subset of the plurality of GNSS satellites onto the image, so that each projected GNSS satellite corresponds to a pixel or a set of coherent pixels of the digital image; determining, for at least a plurality of pixels of the image, a potential GNSS signal quality value; and assigning to each of the projected GNSS satellites the potential GNSS signal quality value of the corresponding pixel or the corresponding set of coherent pixels ([0082], [0077], [0078] and [0032], as quoted in the rejection of claim 5).
Regarding Claim 16, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) teaches: wherein extracting signal features from a GNSS signal comprises considering further information about the GNSS signal, wherein the further information at least comprises a pseudorange ([0085]: “the pseudorange computed for a particular GNSS signal is weighted in accordance with the weight”).
This claim then presents an “and/or” statement; the alternative relied upon is the second, and the enumerated-parameter alternative need not be addressed.
Weisenburger et al. (‘973) teaches: each GNSS signal comprises a pseudorandom noise, and each pseudo-random noise is correlated to obtain the pseudorange ([0034]: “pseudo-random-noise (PRN) codes modulated onto carrier frequencies”; [0070]: “Digital samples 1034 generated by RF front end 1030 may be sent to a correlator 1042, which may perform one or more correlations on digital samples 1034 using local codes”).
Regarding Claim 18, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) teaches: A computer program product comprising program code which is stored on a non-transitory machine-readable medium, and having computer-executable instructions for causing a computer to perform the method according to claim 1 ([0073]: “Method 1100 may be implemented as a computer-readable medium or computer program product comprising instructions which, when the program is executed by one or more computers, cause the one or more computers to carry out the steps of method 1100”).
Regarding Claim 20, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches:
Claim 20 recites the same limitations as claim 1 with the single exception of the image-processing limitation, and is rejected for the reasons set forth in the rejection of claim 1 as to all shared limitations. As to the differing limitation:
Weisenburger et al. (‘973) teaches: processing the image to extract image features, wherein processing the image comprises using image segmentation ([0077]: “At step 1106, the image is segmented into a plurality of regions (e.g., regions 662) based on RF characteristics of objects in the image. Each region of the plurality of regions may be characterized by an extent to which the plurality of GNSS signals are obstructed by the objects within the region”; [0078]: “the image may be segmented such that each pixel of the image is included in one of the plurality of regions”).
Regarding Claim 21, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 20, and further teaches:
The limitations of claim 21 are identical to those of claim 5 and are rejected for the reasons set forth in the rejection of claim 5.
Regarding Claim 22, Weisenburger et al. (‘973) in view of Werner et al. (‘837) teaches the method according to claim 21, and further teaches:
As set forth in the claim objection above, claim 22 is interpreted as depending from claim 21.
Weisenburger et al. (‘973) teaches: deriving the signal classification and/or the estimated local error is also based on the potential GNSS signal quality values ([0084]: “each of the plurality of GNSS signals may be processed differently based on their corresponding region”). The alternative relied upon is the estimated local error, consistent with claim 20.
Weisenburger et al. (‘973) teaches: processing the image comprises identifying at least the second subset of pixels in the image ([0050]: “Region 662-3 corresponds to buildings and includes objects that provide significant RF obstruction”).
Weisenburger et al. (‘973) teaches: determining the potential GNSS signal quality value for a GNSS satellite is based on a relative position of the pixel or the set of coherent pixels corresponding to the respective GNSS satellite relative to the second subset of pixels in the image ([0032]: “the pixel location is compared to the segmented regions to determine in which region the pixel location is located”).
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Weisenburger et al. (US 2022/0018973 A1) in view of Werner et al. (US 2020/0049837 A1), and further in view of Kim et al. (US 2020/0132861 A1).
Regarding Claim 2, Weisenburger et al. (‘973) in view of Werner et al. (‘837) and further in view of Kim et al. (‘861) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) does not explicitly teach, but Kim et al. (‘861) teaches: wherein extracting the signal features comprises applying machine learning to compute a signal feature vector or a signal embedding vector for the respective GNSS signal ([0069]: “There is one encoding subnetwork for each satellite, i.e., for phase measurements from each satellite”; [0076]: “The individual weighted summers weights the features of the encoders with the determined weights”). This element presents an “or” statement; the alternative relied upon is a signal feature vector, and the signal embedding vector alternative need not be addressed. Kim assigns a dedicated trained encoding subnetwork to each satellite, which transforms that satellite’s measurements into a set of encoder features in a hidden layer. Weisenburger obtains its per-satellite pseudorange by deterministic correlation ([0070]) and does not apply a learned model to produce the signal features.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to generate the per-satellite signal features of the combined system of Weisenburger et al. (‘973) and Werner et al. (‘837) using a per-satellite trained encoding subnetwork as taught by Kim et al. (‘861). One would have been motivated to do so because Kim addresses the same multipath-induced measurement corruption as both other references and teaches that a learned per-satellite representation captures behavior that no closed-form model describes (Kim et al. (‘861), [0011]: “there is no deterministic model to” describe the noisy signal, continuing that the relationship is instead “learned through training”), and because Werner already accepts a learned representation of per-satellite measurement quality in place of receiver-computed uncertainty values, making a learned front-end representation a direct extension of the same design approach. There is a reasonable expectation of success because Kim’s encoding subnetworks operate on per-satellite measurements of the same type that Weisenburger’s correlator already produces, and Kim demonstrates that one subnetwork per satellite can be trained and executed within a conventional GNSS receiver processing chain.
Claims 4 and 11 are rejected under 35 U.S.C. 103 as being unpatentable over Weisenburger et al. (US 2022/0018973 A1) in view of Werner et al. (US 2020/0049837 A1), and further in view of Yue Zhe et al. (CN 116088012 A).
Regarding Claim 4, Weisenburger et al. (‘973) in view of Werner et al. (‘837) and further in view of Yue Zhe et al. (‘012) teaches the method according to claim 1, and further teaches:
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: the extracted signal features comprise a signal feature vector or a signal embedding vector ([0064]: “Such parameters may include, but are not limited to: pseudorange; pseudorange uncertainty; range rage; range rate uncertainty; a multipath indicator; elevation above horizon; azimuth”). The alternative relied upon is a signal feature vector; the signal embedding vector alternative need not be addressed. Werner assembles this ordered set of per-satellite parameters as the input to the model, which constitutes a signal feature vector for that satellite.
Weisenburger et al. (‘973) teaches: the extracted image features comprise an image feature vector or an image embedding vector ([0032]: “The identified region is associated with the GNSS satellite and the corresponding GNSS signal”). The alternative relied upon is an image feature vector; the image embedding vector alternative need not be addressed.
Weisenburger et al. (‘973) does not explicitly teach, but Yue Zhe et al. (‘012) teaches: combining the extracted signal features and the extracted image features comprises combining the signal feature vector or the signal embedding vector, respectively, with the image feature vector or the image embedding vector, respectively; and the feature combination is a combined feature vector or a combined embedding vector ([0031]: “the weights of non-line-of-sight propagation signal observations are determined according to the satellite elevation angle, signal-to-noise ratio, and signal occlusion degree”; [0032]: “the signal occlusion degree is characterized by a minimum pixel distance”). Yue Zhe assembles, for each satellite, a common set comprising two signal-derived quantities and one image-derived quantity, which is a combined feature vector for that satellite. Weisenburger applies a scalar weight indexed by region to the pseudorange and does not form a combined vector of signal and image features.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to assemble the per-satellite pseudorange and other signal parameters of the combined system of Weisenburger et al. (‘973) and Werner et al. (‘837) together with the image-derived quantity obtained for that satellite into a single combined per-satellite feature vector, as taught by Yue Zhe et al. (‘012). One would have been motivated to do so because Yue Zhe teaches, in the same sky-facing camera and satellite-projection context as Weisenburger, that a binary in-sky or not-in-sky determination discards usable information and that retaining a graded image-derived measure alongside the signal quantities permits fuller exploitation of obstructed observations rather than their outright rejection (Yue Zhe et al. (‘012), Abstract: “setting different weights according to a weight optimization theory to realize full mining of useful information of different GNSS NLOS signals”). There is a reasonable expectation of success because Werner’s model already accepts an ordered set of per-satellite parameters as its input, so appending the image-derived value taught by Yue Zhe requires only the addition of one input element and no change to the model architecture or to Weisenburger’s projection geometry.
Regarding Claim 11, Weisenburger et al. (‘973) in view of Werner et al. (‘837) and further in view of Yue Zhe et al. (‘012) teaches the method according to claims 1 and 5, and further teaches:
Weisenburger et al. (‘973) does not explicitly teach, but Yue Zhe et al. (‘012) teaches: determining the potential GNSS signal quality value comprises extracting an image feature vector for each GNSS satellite of at least a subset of the projected satellites, the image feature vector comprising the potential GNSS signal quality value ([0033]: “Extract the pixel coordinates of the sky region segmentation edge pixels contained in a circle with the satellite projection position as the center and a radius of R pixels”; [0034]: “Calculate the minimum pixel distance between the satellite projection position and the segmentation edge”). Yue Zhe extracts, for each projected satellite, a quantity computed from the pixels surrounding that satellite’s projection position, and that quantity is the occlusion degree characterizing the signal quality for that satellite. Weisenburger assigns the region value at the single projected pixel location and does not compute a value over a pixel neighborhood defined by the projection position.
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: a signal feature vector is generated for the satellite signal of each GNSS satellite of at least the subset of the projected satellites ([0064]: “the parameters may be used to estimate the position of the electronic device 102 relative to a GNSS satellite (e.g., one of the GNSS satellites 104a-104d) associated with the GNSS receiver data 502”).
Weisenburger et al. (‘973) does not explicitly teach, but Yue Zhe et al. (‘012) teaches: combining the extracted signal features and the extracted image features comprises combining the image feature vector and the signal feature vector of each GNSS satellite of at least the subset of the projected satellites into a combined feature vector for the respective GNSS satellite ([0031]: “the weights of non-line-of-sight propagation signal observations are determined according to the satellite elevation angle, signal-to-noise ratio, and signal occlusion degree”).
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: deriving the signal classification and/or the estimated local error is performed by a classifier module embodied as a neural network or a support vector machine and based on the combined feature vectors of at least the subset of the projected satellites as input ([0057]: “Examples of the algorithms used for training and/or testing the machine learning model 412 include, but are not limited to, linear regression, boosted trees, multi-layer perceptron and/or random forest algorithms”; [0068]: “The machine learning model 412 may be configured to output an amount of error for the GNSS receiver data 502 with respect with the GNSS satellite”). This element presents an “or” statement; the alternative relied upon is a neural network, a multi-layer perceptron being a neural network, and the support vector machine alternative need not be addressed.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to compute, for each satellite projected onto the image in the system of Weisenburger et al. (‘973), a pixel-neighborhood quantity of the kind taught by Yue Zhe et al. (‘012) and to supply that quantity together with that satellite’s signal parameters as the combined input to the model of Werner et al. (‘837). One would have been motivated to do so because Yue Zhe teaches, on the identical segmented sky image with satellites projected by azimuth and elevation, that the distance from a satellite’s projection position to the boundary of the sky region quantifies how deeply that satellite is occluded, information that Weisenburger’s three-way region assignment discards entirely, and because Werner’s model requires graded numerical inputs rather than categorical labels in order to regress a per-satellite error. There is a reasonable expectation of success because Yue Zhe computes the quantity from the same segmented image and the same projected pixel coordinates that Weisenburger already produces, requiring only a distance computation over already-available pixel data.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Weisenburger et al. (US 2022/0018973 A1) in view of Werner et al. (US 2020/0049837 A1) and Yue Zhe et al. (CN 116088012 A), and further in view of Kim et al. (US 2020/0132861 A1).
Regarding Claim 12, Weisenburger et al. (‘973) in view of Werner et al. (‘837) and Yue Zhe et al. (‘012) and further in view of Kim et al. (‘861) teaches the method according to claims 1, 5, and 11, and further teaches:
Weisenburger et al. (‘973) does not explicitly teach, but Kim et al. (‘861) teaches: combined feature vectors of a plurality of points of time (t0, t1, t2) are generated; the classifier module is a recurrent neural network; and for considering a behaviour of the GNSS signals over time, signal classifications and/or estimated local errors at a plurality of points of time (t0, t1, t2) are computed by the classifier module based on the combined feature vectors ([0012]: “some embodiments train a recurrent neural network (RNN) to determine a position of the vehicle from a set of phase measurements in a presence of noise caused by a multipath transmission of at least some of the satellite signals at some instances of time”; Abstract: “track the position of the vehicle over different instances of time by processing the set of phase measurements received at each instance of time with the recurrent neural network”). The alternative relied upon is estimated local errors, consistent with claim 1. Neither Weisenburger nor Werner processes a temporal sequence of per-satellite inputs with a recurrent architecture.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to implement the classifier module of the combined system as a recurrent neural network operating on the per-satellite combined feature vectors at successive epochs, as taught by Kim et al. (‘861). One would have been motivated to do so because Kim teaches that multipath corruption of GNSS measurements exhibits temporal structure that a feedforward model cannot capture, and that a recurrent architecture retains internal state across epochs so as to exploit that structure (Kim et al. (‘861), [0012]: “Unlike feedforward neural networks, RNNs can use their internal state (memory) to process sequences of inputs”), and because in the combined system the camera and the satellites both move continuously so that each satellite’s occlusion condition evolves across epochs in a manner that a single-epoch evaluation cannot represent. There is a reasonable expectation of success because Kim applies the recurrent architecture to per-satellite GNSS measurements corrupted by the same multipath mechanism at issue in Weisenburger and Werner, and the substitution requires only that the existing per-satellite feature vectors be buffered across epochs before being supplied to the model.
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Weisenburger et al. (US 2022/0018973 A1) in view of Werner et al. (US 2020/0049837 A1), and further in view of Siercks et al. (US 2020/0240784 A1).
Regarding Claim 17, Weisenburger et al. (‘973) in view of Werner et al. (‘837) and further in view of Siercks et al. (‘784) teaches:
Weisenburger et al. (‘973) teaches: A system for processing satellite signals to derive a geospatial position, the system comprising: a GNSS antenna configured to receive GNSS signals from a plurality of GNSS satellites ([0075]: “The plurality of GNSS signals are received via a GNSS antenna (e.g., GNSS antennas 116, 216, 316, 416, 516, 916)”).
Weisenburger et al. (‘973) teaches: a measurement engine configured to correlate pseudo-random noises of the GNSS signals to obtain pseudoranges ([0070]: “Digital samples 1034 generated by RF front end 1030 may be sent to a correlator 1042, which may perform one or more correlations on digital samples 1034 using local codes”).
Weisenburger et al. (‘973) teaches: a positioning engine configured to compute a geospatial position based on a subset of the GNSS signals and/or pseudoranges ([0071]: “Based on multiple pseudoranges corresponding to multiple GNSS satellites 902… receiver processor 1036 may generate and output position data 1038 comprising a plurality of GNSS points”). This element presents an “and/or” statement; the alternative relied upon is pseudoranges, and the GNSS signals alternative need not be addressed.
Weisenburger et al. (‘973) does not explicitly teach, but Siercks et al. (‘784) teaches: a SLAM unit configured to determine an orientation of the imaging device while capturing the image ([0103]: “a series of images of the surrounding is captured with the at least one camera, the series comprising an amount of images captured with different poses of the camera, the poses representing respective positions and orientations of the camera”; [0104]: “a SLAM-evaluation with a defined algorithm using the series of images is performed… and the poses for the images are determined”). The optional recitation particularly using visual SLAM based on image data captured by the imaging device is not limiting under MPEP § 2111.04 but is noted as taught, Siercks performing the SLAM evaluation on the captured image series itself. Weisenburger determines image orientation from an orientation sensor or from GNSS data ([0079]) rather than by a SLAM unit.
It would have been obvious to a PHOSITA before the effective filing date of the claimed invention to determine the orientation of the imaging device in the system of Weisenburger et al. (‘973) by means of a SLAM unit operating on the captured image series, as taught by Siercks et al. (‘784). One would have been motivated to do so because Siercks addresses the identical hardware configuration of a camera rigidly mounted with a GNSS antenna on a surveying pole and teaches that SLAM evaluation of the image series yields the camera poses, including orientation, directly from the imagery already being captured, eliminating dependence on an inertial sensor whose bias Siercks elsewhere identifies as requiring calibration and compensation ([0154]: the inertial measuring unit is calibrated “wherein a systematic error of the inertial measuring unit is compensated”). Because Weisenburger expressly conditions the accuracy of its satellite projection on accurate orientation measurement ([0033]: “Projecting the orbital positions of the GNSS satellites is facilitated by an accurate measurement of the position”), a PHOSITA would have had reason to adopt a drift-free image-based orientation source. There is a reasonable expectation of success because Siercks implements the SLAM evaluation on a camera module attached to a surveying pole carrying a GNSS antenna, the same platform geometry disclosed by Weisenburger, and the evaluation consumes the image series that Weisenburger’s imaging device already produces.
Weisenburger et al. (‘973) teaches: a signal-processing module configured to extract signal features from the GNSS signals ([0070]: “Correlator 1042 may generate correlation results 1046 based on digital samples 1034 and control parameters 1044 and send these results to receiver processor 1036”).
Weisenburger et al. (‘973) teaches: an image-processing module configured to process the image to extract image features by using image segmentation, and/or by applying machine learning to compute an image feature vector or an image embedding vector for at least a third subset of pixels of the image, the third subset of pixels being defined based on a position of the respective GNSS satellite ([0077]: “At step 1106, the image is segmented into a plurality of regions (e.g., regions 662) based on RF characteristics of objects in the image”). This element presents an “and/or” statement; the alternative relied upon is using image segmentation, and the machine-learning alternative need not be addressed.
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: a classifier module embodied as a neural network or a support vector machine and configured to derive, based on feature combinations of a plurality of GNSS satellites and by applying machine learning, a signal classification and/or an estimated local error of the respective GNSS signal, each feature combination being a combination of the extracted signal features and the extracted image features of the respective GNSS signal ([0057]: “multi-layer perceptron and/or random forest algorithms”; [0068]: “The machine learning model 412 may be configured to output an amount of error for the GNSS receiver data 502 with respect with the GNSS satellite”). The alternatives relied upon are a neural network and an estimated local error. The obviousness rationale, motivation to combine and reasonable expectation of success set forth under claim 1 apply equally to this structural element and are expressly incorporated.
Weisenburger et al. (‘973) does not explicitly teach, but Werner et al. (‘837) teaches: wherein the positioning engine is configured to compute the geospatial position based also on the signal classification and/or on the estimated local error for each GNSS signal of the subset ([0069]: “the location estimator 500 may disregard and/or limit measurements corresponding to that GNSS satellite”). The obviousness rationale, motivation to combine and reasonable expectation of success set forth under claim 1 apply equally to this structural element and are expressly incorporated.
Allowable Subject Matter
The following is a statement of reasons for the indication of allowable subject matter:
Claims 7, 8, and 23 are objected to as being dependent upon a rejected base claim but would be allowable if rewritten in independent form including all the limitations of the base claim and any intervening claims.
No reference of record teaches centring, for at least a subset of the plurality of GNSS satellites, the image on one of the subset of GNSS satellites based on the orientation of the image and on known satellite positions, nor assigning to each of the GNSS satellites of the subset the potential GNSS signal quality value of the pixel or a set of coherent pixels at the centre of the image. Weisenburger et al. (‘973) and Yue Zhe et al. (‘012) both project satellites onto a fixed image frame and read the image value at the projected location; neither reorients the image per satellite so that the satellite occupies the image centre. Claim 8 depends from claim 7 and claim 23 incorporates the same centring limitations.
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