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 .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 6/2/2026 has been entered.
Response to Amendment
This Office Action is in response to the RCE filed on 6/2/2026. Claims 1, 10, and 21 are amended. Claims 1-17 and 20-35 are currently pending.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-3, 10-12, 14-17, 21-23, 26-30, and 32-35 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by US 20190325595 A1 hereinafter Stein.
Regarding claim 1, Stein teaches a method comprising (a method for vehicle environment modeling with a camera paragraph [0030]): obtaining one or more vehicle motion profile applied to a portion of one or more vehicles traversing a first road segment, (FIGS. 7-8, data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076]), wherein the one or more vehicle motion profiles are obtained based at least partially on measurements of motion with one or more inertial sensors disposed on a portion of the one or more vehicles; and (The system 208 includes processing circuitry to perform vehicle environment modeling via images obtained from the camera 202. The vehicle environment modeling may include modeling the road surface 206, obstacles, obstructions, and moving bodies paragraph [0044] The inertial sensors may be integrated in the camera, or may be mounted elsewhere on or in the vehicle. Rotational estimates may instead, or additionally, be obtained computationally from one or more previous image frames. Paragraph [0131]).;
Training a statistical model using the one or more vehicle motion profiles (FIGS. 7-8 illustrates an example of a DNN training system, according to an embodiment. Here, a multi-modal loss function application engine 950 is configured to supply training data 930 as input to DNN Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example. paragraph [0076]), the trained statistical model being configured when trained to identify or classify one or more road features associated with a second road segment (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076]) based at least in part on the vehicle motion profiles applied to a portion of a second vehicle traversing a second road segment; (The system 208 is arranged to provide a data set to an artificial neural network (ANN) to produce a gamma image paragraph [0045], the system is used to accurately estimate the planar (or bi-quadratic) model of the road surface, and compute small deviations from the planar (or bi-quadratic) road surface model to detect or quantify various surface features 104 paragraph [0190])
Regarding claim 2, Stein discloses the method of claim 33, further comprising:
associating the one or more road features of the second road segment with one or more geographical locations; and (the target location and size are inputted as images. Target location includes two gradient images in which pixel values represent a distance from the center of the target. Here, a horizontal gradient image 1106 (e.g., position x or P.sub.x) and a vertical gradient image 1108 (e.g., position y or P.sub.y) make up the target location paragraph [0097])
storing, in non-volatile computer readable memory, the one or more geographical locations of the one or more road features of the second road segment. (Registers of the processor 2702, the main memory 2704, the static memory 2706, or the mass storage 2708 may be, or include, a machine readable medium 2722 on which is stored one or more sets of data structures or instructions 2724 (e.g., software) embodying or utilized by any one or more of the techniques or functions described herein [paragraph [0242], Specific examples of non-transitory machine readable media may include: non-volatile memory, such as semiconductor memory devices (e.g., Electrically Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM)) and flash memory devices; magnetic disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks paragraph [0243])
The examiner notes in this instance that the geographical locations are data structures.
Regarding claim 3, Stein discloses the method of claim 2, further comprising generating a map based on the one or more geographical locations. (Each layer produces a feature map, which is in turn passed to the subsequent layer for further processing along forward propagation path 508. As depicted, the operations of convolutional network portion 502 operate to progressively reduce resolution of the feature maps, while increasing the number of channels (dimensionality) of the feature maps along convolutional forward propagation path 508A. The operations of deconvolutional network portion 504 operate to progressively increase resolution of the feature maps, while decreasing their dimensionality along deconvolutional forward propagation path 508B paragraph [0067], Operation of the DNN in inference mode produces a road structure map such as a gamma map as described above. Paragraph [0108])
Regarding claim 10, Stein discloses the method of claim 33, wherein obtaining the vehicle motion profile caused by the one or more road features of the second road segment comprises traversing the second road segment with the one or more vehicles while measuring the vertical motion of the portion of the one or more vehicles using one or more motion sensors disposed in the one or more vehicles. (The system 208 is arranged to model the road surface 206 using the gamma image. In an example, modeling the road surface includes computing a vertical deviation from the plane of a road surface feature paragraph [0054])
Regarding claim 11, Stein discloses the method of claim 10, wherein the portion of the one or more vehicles includes a wheel of the one or more vehicles. (In another an example, where the suspension state of the vehicle is available, suspension information is considered together with the ego-motion to more accurately measure the vertical motion of the vehicle's wheel paragraph [0090])
Regarding claim 12, Stein discloses the method of claim 10, wherein the vehicle motion profile caused by the one or more road features of the second road segment is measured as a function of time. (To perform the modeling, the system 208 is arranged to obtain a time-ordered sequence of images representative of the road surface 206 paragraph [0044])
Regarding claim 14, Stein discloses the method of claim 33, wherein the trained statistical model is a first trained statistical model (The system 208 is arranged to provide a data set to an artificial neural network (ANN) to produce a gamma image paragraph [0045]), further comprising:
training a second trained statistical model using the vehicle motion profile caused by the one or more road features of the first road segment, (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076], invoking a second ANN using the three-dimensional structure to determine whether the features represent an object moving or not moving within an environment of the road surface paragraph [0294]) wherein the second trained statistical model is configured to identify or classify one or more road feature characteristics based at least in part on the vehicle motion profile caused by the one or more road features of the second road segment; and (wherein the ANN and the second ANN are implemented as a single ANN trained to produce a two-channel output, wherein a first channel is the three-dimensional structure of the scene and the second channel is the three-dimensional structure produced by the second ANN that used using more photogrammetric loss in its training paragraph [0293])
using the second trained statistical model to classify or identify one or more road features of the second road segment based at least partly on measurements of a vehicle motion profile caused by the one or more road features of the second road segment. (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076] The system 208 is arranged to provide a data set to an artificial neural network (ANN) to produce a gamma image paragraph [0045], the system is used to accurately estimate the planar (or bi-quadratic) model of the road surface, and compute small deviations from the planar (or bi-quadratic) road surface model to detect or quantify various surface features 104 paragraph [0190])
Regarding claim 15, Stein discloses the method of claim 14, wherein to classify or identify one or more road features of the second road segment includes identifying a road feature type or road feature characteristics. (invoking a second ANN using the three-dimensional structure to determine whether the features represent an object moving or not moving within an environment of the road surface paragraph [0294])
Regarding claim 16, Stein discloses the method of claim 15, wherein the road feature type includes one selected from a group of a speed bump, a pothole, a manhole cover, a storm grate, a frost heave, and an expansion joint. (Using systems and methods provided herein, surface features such as bumps or holes, speed bumps, curbs, or manhole covers, may be measured or modeled as vertical deviations from the road surface (e.g., plane) with sub-pixel accuracy (e.g., on the order of 1-2 centimeters paragraph [0191])
Regarding claim 17, Stein discloses the method of claim 15, wherein the one or more road feature characteristics include a size of the one or more road features. (Here, a horizontal gradient image 1106 (e.g., position x or P.sub.x) and a vertical gradient image 1108 (e.g., position y or P.sub.y) make up the target location input to the neural network 1112. These images include an outline of the target to illustrate the gradient's relationship to the target. The target size is represented here as an image in which all of the pixels have the same value (e.g., a constant value image) representative of the target's size paragraph [0097])
Regarding claim 21, Stein discloses a method comprising:
obtaining first vehicle motion profiles applied to a portion of one or more vehicles traversing a first road segment associated with one or more road features, wherein the first vehicle motion profiles are obtained based at least partially on measurements of motion with one or more inertial sensors disposed on a portion of the one or more vehicles traversing the first road segment; (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076] Paragraph [0131])
obtaining second vehicle motion profiles applied to a portion of one or more vehicles traversing a second road segment associated with an absence of the one or more road features, wherein the second vehicle motion profiles are obtained based at least partially on measurements of motion with one or more inertial sensors disposed on a portion of the one or more vehicles traversing the second road segment; (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076] The inertial sensors may be integrated in the camera, or may be mounted elsewhere on or in the vehicle. Rotational estimates may instead, or additionally, be obtained computationally from one or more previous image frames. Paragraph [0131])
generating a trained statistical model using the first vehicle motion profiles and the second vehicle motion profiles; and (FIGS. 7-8 illustrates an example of a DNN training system, according to an embodiment. Here, a multi-modal loss function application engine 950 is configured to supply training data 930 as input to DNN Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example. paragraph [0076])
storing in non-volatile computer readable memory, the trained statistical model. (Registers of the processor 2702, the main memory 2704, the static memory 2706, or the mass storage 2708 may be, or include, a machine readable medium 2722 on which is stored one or more sets of data structures or instructions 2724 (e.g., software) embodying or utilized by any one or more of the techniques or functions described herein [paragraph [0242], Specific examples of non-transitory machine readable media may include: non-volatile memory, such as semiconductor memory devices (e.g., Electrically Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM)) and flash memory devices; magnetic disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks paragraph [0243])
The examiner notes that the trained statistical model in this instance is software.
Regarding claim 22, Stein discloses the method of claim 21, wherein the trained statistical model is a first trained statistical model, wherein the method further comprises: (The system 208 is arranged to provide a data set to an artificial neural network (ANN) to produce a gamma image paragraph [0045])
obtaining third vehicle motion profiles applied to a portion of the one or more vehicles traversing a first type of road feature; (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076])
obtaining road feature characteristic data associated with the third vehicle motion profiles; (using the three-dimensional structure to determine whether the features represent an object moving or not moving within an environment of the road surface paragraph [0294])
generating a second trained statistical model using the third vehicle motion profiles and the road feature characteristic data; and (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076], invoking a second ANN using the three-dimensional structure to determine whether the features represent an object moving or not moving within an environment of the road surface paragraph [0294])
storing in the non-volatile computer readable memory, the second trained statistical model. (Registers of the processor 2702, the main memory 2704, the static memory 2706, or the mass storage 2708 may be, or include, a machine readable medium 2722 on which is stored one or more sets of data structures or instructions 2724 (e.g., software) embodying or utilized by any one or more of the techniques or functions described herein [paragraph [0242], Specific examples of non-transitory machine readable media may include: non-volatile memory, such as semiconductor memory devices (e.g., Electrically Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM)) and flash memory devices; magnetic disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks paragraph [0243])
The examiner notes that the trained statistical model in this instance is software.
Regarding claim 23, Stein discloses the method of claim 22, wherein the first type of road feature includes one selected from a group of a speed bump, a pothole, a manhole cover, a storm grate, a frost heave, and an expansion joint. (Using systems and methods provided herein, surface features such as bumps or holes, speed bumps, curbs, or manhole covers, may be measured or modeled as vertical deviations from the road surface (e.g., plane) with sub-pixel accuracy (e.g., on the order of 1-2 centimeters paragraph [0191])
Regarding claim 26, Stein discloses the method of claim 21, wherein obtaining the first vehicle motion profile comprises:
traversing, in the one or more vehicles, a first plurality of road segments, wherein each road segment of the first plurality of road segments includes the one or more road features; and (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076])
while traversing each road segment of the first plurality of road segments, measuring vehicle motion of the portion of the one or more vehicles. (In an example, the ego-motion may be provided by an ego-motion sensor and processing engine. This type of engine uses robust tracking of points on the road and the points above the road using an essential matrix. It also combines any inertial sensors and speedometer information available paragraph [0142])
Regarding claim 27, Stein discloses the method of claim 21, wherein obtaining the second vehicle motion profile comprises:
traversing, in the one or more vehicles, a second plurality of road segments, wherein each road segment of the second plurality of road segments does not include the one or more road features; and (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076])
while traversing each road segment of the second plurality of road segments, measuring vehicle motion of the portion of the one or more vehicles. (In an example, the ego-motion may be provided by an ego-motion sensor and processing engine. This type of engine uses robust tracking of points on the road and the points above the road using an essential matrix. It also combines any inertial sensors and speedometer information available paragraph [0142])
Regarding claim 28, Stein discloses the method of claim 26, wherein the measured vehicle motion includes a vertical motion of the portion of the one or more vehicles. (The system 208 is arranged to model the road surface 206 using the gamma image. In an example, modeling the road surface includes computing a vertical deviation from the plane of a road surface feature paragraph [0054])
Regarding claim 29, Stein discloses the method of claim 26, wherein the measured vehicle motion includes a longitudinal motion of the one or more vehicles. (The epipole is a vector that represents the direction of forward motion. In an example, image-formatted epipole location data 2578 includes a pair of images, each image having a resolution that is the same or similar to image frames A, B, and C paragraph [0229])
Regarding claim 30, Stein discloses the method of claim 27, wherein the portion of the one or more vehicles includes a wheel. (In another an example, where the suspension state of the vehicle is available, suspension information is considered together with the ego-motion to more accurately measure the vertical motion of the vehicle's wheel paragraph [0090])
Regarding claim 32, Stein discloses at least one non-transitory computer-readable storage medium storing programming instructions that, when executed by at least one processor, causes the at least one processor to perform the method of any one of claim 1. (In an example, the hardware of the circuitry may include variably connected physical components (e.g., execution units, transistors, simple circuits, etc.) including a machine readable medium physically modified (e.g., magnetically, electrically, moveable placement of invariant massed particles, etc.) to encode instructions of the specific operation. Paragraph [0239], Non-limiting machine readable medium examples may include solid-state memories, optical media, magnetic media, and signals (e.g., radio frequency signals, other photon based signals, sound signals, etc.). In an example, a non-transitory machine readable medium comprises a machine readable medium with a plurality of particles having invariant (e.g., rest) mass, and thus are compositions of matter. Accordingly, non-transitory machine-readable media are machine readable media that do not include transitory propagating signals. Specific examples of non-transitory machine readable media may include: non-volatile memory paragraph [0243], FIG. 27)
Regarding claim 33, Stein discloses the method of claim 1. Stein additionally discloses further comprising using the trained statistical model to identify or classify one or more road features (FIGS. 7-8, Training data 830 may include various sequences of image frames captured by one or more vehicle-mounted cameras. The image frames may include video footage captured on various roads, in various geographic locales, under various lighting and weather conditions, for example paragraph [0076]) of the second road segment based at least in part on measurements of a vehicle motion profile applied to a portion of a vehicle traversing the second road segment. (The system 208 is arranged to provide a data set to an artificial neural network (ANN) to produce a gamma image paragraph [0045], the system is used to accurately estimate the planar (or bi-quadratic) model of the road surface, and compute small deviations from the planar (or bi-quadratic) road surface model to detect or quantify various surface features 104 paragraph [0190])
Regarding claim 34, Stein discloses the method of claim 1. Stein additionally discloses wherein the trained statistical model is configured, when trained, to identify and classify one or more road features associated with the second road segment based at least in part on the motion profiles. (The system 208 is arranged to provide a data set to an artificial neural network (ANN) to produce a gamma image paragraph [0045], the system is used to accurately estimate the planar (or bi-quadratic) model of the road surface, and compute small deviations from the planar (or bi-quadratic) road surface model to detect or quantify various surface features 104 paragraph [0190])
Regarding claim 35, Stein discloses the method of claim 1. Stein additionally discloses wherein the first road segment and the second road segment are one road segment. (At operation 1501, a sequence of image frames (e.g., a first image frame A, a second image frame B, and a third image frame C) of the same portion of a road in field of view of a camera are captured. Image points of the road in first image frame A are matched at operation 1502 to corresponding image points of the road in the second image frame B. Likewise, image points of the road in the second image frame B are matched at operation 1502 to corresponding image points of the road in the third image frame C. Paragraph [0122])
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.
Claim(s) 4-8, 13 20, 24, 25, and 31 are rejected under 35 U.S.C. 103 as being unpatentable over Stein in view of US 20190324443 A1 hereinafter Cella.
Regarding claim 4, Stein discloses the method as recited in claim 33. However, Stein doesn’t explicitly disclose filtering the vehicle motion profile caused by by the one or more road features of the second road segment to attenuate one or more vehicle-specific characteristics prior to inputting the vehicle motion profile caused by the one or more road features of the second road segment to the trained statistical model.
Cella discloses filtering the vehicle motion profile caused by by the one or more road features of the second road segment to attenuate one or more vehicle-specific characteristics prior to inputting the vehicle motion profile caused by the one or more road features of the second road segment to the trained statistical model. (a network control circuit 11710 for sending and receiving information related to the sensor inputs to an external system and a data filter circuit configured to dynamically adjust what portion of the information is sent based on instructions received over the network communication interface paragraph [1268])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the filtering circuit of Cella to the vehicle motion profile of Stein prior to the inputting of the profile to the trained statistical model. This combination would enable the trained statistical model to generate more accurate road features by leveraging high-confidence input data.
Regarding claim 5, Stein and Cella disclose the method as recited in claim 4. However, Stein doesn’t explicitly disclose where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises filtering a first frequency of the vehicle motion profile caused by the one or more road features of the second road segment to reduce artifacts of wheel-hop from the vehicle motion profile caused by the one or more road features of the second road segment.
Cella discloses where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises filtering a first frequency of the vehicle motion profile caused by the one or more road features of the second road segment to reduce artifacts of wheel-hop from the vehicle motion profile caused by the one or more road features of the second road segment. (a band pass filter circuit 8532 which may be used to separate out signals occurring at different frequencies paragraph [0658], In embodiments, the sensor inputs additionally comprise microphones or vibration sensors configured to detect vibrational or audio-frequency conditions in movable or rotational components, such as whirring, howling, growling, whining, rumbling, clunking, rattling, wheel hopping, and chattering paragraph [1310])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the filtering frequency of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to reduce the noise / inaccurate sensor data that would be obtained via wheel hopping thus making the data more accurate.
Regarding claim 6, Stein and Cella disclose the method as recited in claim 5. However, Stein doesn’t explicitly disclose where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises applying a notch filter to the vehicle motion profile caused by the one or more road features of the second road segment, wherein a stop-band frequency range of the notch filter includes a frequency of the wheel-hop.
Cella discloses where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises applying a notch filter to the vehicle motion profile caused by the one or more road features of the second road segment, wherein a stop-band frequency range of the notch filter includes a frequency of the wheel-hop. (Additionally, or alternatively, a band pass filter circuit 8532 includes one or more notch filters or other filtering mechanism to narrow ranges of frequencies (e.g., frequencies from a known source of noise). This may be used to filter out dominant frequency signals such as the overall rotation, and may help enable the evaluation of low amplitude signals at frequencies associated with torsion, bearing failure and the like paragraph 0658, In embodiments, the sensor inputs additionally comprise microphones or vibration sensors configured to detect vibrational or audio-frequency conditions in movable or rotational components, such as whirring, howling, growling, whining, rumbling, clunking, rattling, wheel hopping, and chattering paragraph [1310])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the notch filter of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to reduce the noise / inaccurate sensor data that would be obtained via wheel hopping thus making the data more accurate.
Regarding claim 7, Stein and Cella disclose the method as recited in claim 5. However, Stein doesn’t explicitly disclose where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises: applying a low-pass filter to the vehicle motion profile caused by the one or more road features of the second road segment, wherein a cutoff frequency of the low-pass filter is less than a frequency of the wheel-hop.
Cella discloses where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises: applying a low-pass filter to the vehicle motion profile caused by the one or more road features of the second road segment, wherein a cutoff frequency of the low-pass filter is less than a frequency of the wheel-hop. (An example band pass filter circuit 8532 includes any filtering operations understood in the art, including at least a low-pass filter paragraph [0658], In embodiments, the sensor inputs additionally comprise microphones or vibration sensors configured to detect vibrational or audio-frequency conditions in movable or rotational components, such as whirring, howling, growling, whining, rumbling, clunking, rattling, wheel hopping, and chattering paragraph [1310])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the low pass filter of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to reduce the noise / inaccurate sensor data that would be obtained via wheel hopping thus making the data more accurate.
Regarding claim 8, Stein and Cella disclose the method as recited in claim 5. However, Stein doesn’t explicitly disclose where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises applying a high-pass filter to the vehicle motion profile caused by the one or more features of the second road segment, wherein a cutoff frequency of the high-pass filter is above a frequency of the wheel-hop.
Cella discloses where filtering the vehicle motion profile caused by the one or more road features of the second road segment comprises applying a high-pass filter to the vehicle motion profile caused by the one or more features of the second road segment, wherein a cutoff frequency of the high-pass filter is above a frequency of the wheel-hop. (An example band pass filter circuit 8532 includes any filtering operations understood in the art, including at least a low-pass filter, a high-pass filter, and/or a band pass filter—for example to exclude or reduce frequencies that are not of interest for a particular determination paragraph [0658], In embodiments, the sensor inputs additionally comprise microphones or vibration sensors configured to detect vibrational or audio-frequency conditions in movable or rotational components, such as whirring, howling, growling, whining, rumbling, clunking, rattling, wheel hopping, and chattering paragraph [1310])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the low pass filter of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to reduce the noise / inaccurate sensor data that would be obtained via wheel hopping thus making the data more accurate.
Regarding claim 13, Stein discloses the method as described in claim 12. However, Stein does not disclose transforming the vehicle motion profile caused by the one or more road features of the second road segment from a time domain into a distance domain prior to inputting the vehicle motion profile caused by the one or more road features of the second road segment to the trained statistical model.
Cella discloses transforming the vehicle motion profile caused by the one or more road features of the second road segment from a time domain into a distance domain prior to inputting the vehicle motion profile caused by the one or more road features of the second road segment to the trained statistical model.
(The signal evaluation circuit 9208 may process the detection values to obtain information about a bearing being monitored. The frequency transformation circuit 9212 may transform one or more time-based detection values to frequency information. The transformation may be accomplished using techniques such as a digital Fast Fourier transform (“FFT”), Laplace transform, Z-transform, wavelet transform, other frequency domain transform, or other digital or analog signal analysis techniques, including, without limitation, complex analysis, including complex phase evolution analysis. [0780])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the transformation technique from a time domain into the distance domain of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to interpret data in terms of physical space rather than just frequency components which would increase the localization of features.
Regarding claim 20, Stein discloses the method as described in claim 33. However, Stein does not disclose wherein the one or more road features correspond to one or more clusters identified in a training data set.
Cella discloses wherein the one or more road features correspond to one or more clusters identified in a training data set. (For example, some of the analysis techniques used in unsupervised learning may include K-means clustering, Gaussian mixture models, Hidden Markov models, and the like. The algorithms used in supervised and unsupervised learning methods of pattern recognition enable the use of pattern recognition in various high precision applications paragraph [0340])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the clustering technique of Cella to the statistical model of Stein. This would enable the system of Stein to improve road feature classification accuracy by automatically segmenting different road types.
Regarding claim 24, Stein discloses the method as described in claim 21. However, Stein does not disclose transforming the first vehicle motion profiles and the second vehicle motion profiles into a frequency domain prior to generating the trained statistical model.
Cella discloses transforming the first vehicle motion profiles and the second vehicle motion profiles into a frequency domain prior to generating the trained statistical model. (The frequency transformation circuit 9212 may transform one or more time-based detection values to frequency information paragraph [0780])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the frequency domain change of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to cherry pick data from signals at frequencies of a specific interest thus increasing the overall efficiency of the system.
Regarding claim 25, Stein discloses the method as described in claim 21. However, Stein does not disclose transforming the first vehicle motion profiles and the second vehicle motion profiles from a time domain into a distance domain prior to generating the trained statistical model.
Cella discloses transforming the first vehicle motion profiles and the second vehicle motion profiles from a time domain into a distance domain prior to generating the trained statistical model. (The signal evaluation circuit 9208 may process the detection values to obtain information about a bearing being monitored. The frequency transformation circuit 9212 may transform one or more time-based detection values to frequency information. The transformation may be accomplished using techniques such as a digital Fast Fourier transform (“FFT”), Laplace transform, Z-transform, wavelet transform, other frequency domain transform, or other digital or analog signal analysis techniques, including, without limitation, complex analysis, including complex phase evolution analysis. [0780])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the transformation technique from a time domain into the distance domain of Cella to the vehicle motion profile of Stein. This combination would enable the system of Stein to interpret data in terms of physical space rather than just frequency components which would increase the localization of features.
Regarding claim 31, Stein discloses the method as described in claim 27. However, Stein does not disclose identifying one or more clusters within the first vehicle motion profiles, wherein the trained statistical model is generated using the one or more clusters.
Cella discloses identifying one or more clusters within the first vehicle motion profiles, wherein the trained statistical model is generated using the one or more clusters. (For example, some of the analysis techniques used in unsupervised learning may include K-means clustering, Gaussian mixture models, Hidden Markov models, and the like. The algorithms used in supervised and unsupervised learning methods of pattern recognition enable the use of pattern recognition in various high precision applications paragraph [0340])
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the clustering technique of Cella to the statistical model of Stein. This would enable the system of Stein to improve road feature classification accuracy by automatically segmenting different road types.
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Stein in view of Cella, and further in view of US 20050046137 A1 hereinafter Dreff.
Regarding claim 9, Stein and Cella disclose the method as described in claim 6. However, Stein and Cella do not explicitly disclose where the frequency of wheel-hop is between 10 and 15 Hz.
Dreff discloses where the frequency of wheel-hop is between 10 and 15 Hz. (The shock absorber is used to attenuate both the low frequency ride modes, which are generally at frequencies less than 2 Hz, and the higher frequency wheel hop and tramp modes, which are typically in the range of 10-15 Hz paragraph [0003])
Response to Arguments
Applicants’ arguments filed 6/2/2026 have been fully considered.
Applicants’ argument “Without acceding to the propriety of the rejection, amend3ed claim 1 recites "one or more vehicle motion profiles . .. obtained based at least partially on measurements of motion with one or more inertial sensors disposed on a portion of the one or more vehicles." Stein, by contrast, is a camera-based system. Stein's training and inference data are image sequences obtained from a vehicle-mounted camera, and "ego-motion" in Stein refers to motion of the camera or vehicle reconstructed from those images, not measurements made by an inertial sensor at a portion of the vehicle. (Stein 11 [0044]-[0049], [0076].) A camera is not an inertial sensor given the plain and ordinary meaning of this term used in the art and in view of the current disclosure. Thus, Stein's image-derived ego-motion is not a measurement of motion by an inertial sensor and cannot disclose any such an inertial sensor being a source of the data used to train or operate a model. Applicant respectfully notes Stein's single reference to suspension state in paragraph [0090] does not cure the deficiency. There, Stein states only that "where the suspension state of the vehicle is available, suspension information is considered together with the ego-motion to more accurately measure the vertical motion of the vehicle's wheel." Stein neither identifies the source of any such suspension-state information as an inertial sensor nor discloses obtaining a vehicle motion profile from such measurements. Suspension state in Stein is at most an optional refinement to image- derived ego-motion used in computing a geometric loss; it is not a sensor-measured motion profile used to train the model as claimed.” Filed 6/2/2026 have been fully considered but are not persuasive.
Applicant contends that Stein fails to disclose obtaining one or more vehicle motion profiles based at least partially on measurements of motion with one or more inertial sensors disposed on a portion of the one or more vehicles because Stein is allegedly a camera based system who’s ego-motion reconstruction is reconstructed solely from image sequences. However, Stein expressly teaches that vehicle motion compensation may be achieved based on rotational estimates or measurements provided by inertial sensors, and further teaches that such inertial sensors may be integrated in the camera or mounted elsewhere on or in the vehicle. Additionally, Stein teaches that the data set supplied to the artificial neural network includes not only a sequence of images but also motion of the sensor. Accordingly, Stein discloses obtaining vehicle motion measurements from inertial sensors disposed on the vehicle.
Applicants’ argument “Second, Stein does not train a statistical model using vehicle motion profiles. Claim 1 recites "training a statistical model using the one or more vehicle motion profiles." Stein's training data, as described in paragraph [0076], is "various sequences of image frames captured by one or more vehicle-mounted cameras." The Office Action's citation to paragraph [0076] for the training limitation conflates image sequences with motion profiles. They are not the same. Stein does not disclose training a model on inertial-sensor-derived motion profiles of a vehicle portion." Filed 6/2/2026 has been fully considered but is not persuasive.
Applicant asserts that the Office Action improperly equates image sequences with vehicle motion profiles. However, Stein is not limited to image frames alone. Stein teaches that the dataset provided to the artificial neural network includes motion of the sensor in addition to image sequences. Further, Stein expressly teaches that rotation estimates or measurements may be provided by inertial sensors integrated into the camera or mounted elsewhere on the vehicle. Accordingly, the training data includes vehicle motion information obtained from the vehicle and not merely image sequences.
Applicants’ argument “Third, Stein's trained model does not identify or classify road features based on motion profiles of a second vehicle traversing a second road segment. Claim 1 recites that the trained model, when trained, will "identify or classify one or more road features associated with a second road segment based at least in part on motion profiles applied to a portion of a second vehicle traversing a second road segment." Stein's trained model produces a "gamma map," which is a three-dimensional reconstruction of the scene, from image input. (Stein [ 0045], [0067], [0108].) Identification of features such as bumps or potholes in Stein is a downstream geometric computation performed on the gamma map, not an output of the trained model in response to a vehicle motion profile input. (Stein [0190]-[0191].) Stein does not disclose inputting a vehicle motion profile of a second vehicle traversing a second road segment to the trained model to identify or classify road features.” Filed 9/11/2025 has been fully considered but is not persuasive.
Applicant asserts that Stein teaches that the trained model merely produces a gamma map and that road feature identification occurs only as a downstream geometric computation. However, Stein teaches that the artificial neural network is trained using road data obtained from vehicle mounted sensors to generate a representation of the road surface that is subsequently used to detect and quantify road surface features, including deviations from the road surface model. The language as claimed requires that the trained statistical model be configured, when trained, to identify to classify one or more road features associated with a road segment. The claim does not require the statistical model to directly output the road features without subsequent processing. Steins trained model generates the learned road representation that enables identification of the road features.
Applicants’ argument “Third, independent claim 21 recites "obtaining first vehicle motion profiles applied to a portion of one or more vehicles traversing a first road segment associated with one or more road features" and "obtaining second vehicle motion profiles applied to a portion of one or more vehicles traversing a second road segment associated with an absence of the one or more road features." The first profiles are associated with a road segment including one or more road features, and the second profiles are associated with a segment having an absence of such features. Stein discloses no such comparative dataset. As discussed above in reference to claim 1, Stein does not disclose collecting any inertial sensor-derived motion profiles of vehicle portions, much less contrasting datasets of profiles with and without features.
The Examiner has argued that Stein inherently produces first and second profiles because roads traversed at varying conditions will naturally include both feature-present and feature-absent segments. Even accepting this characterization, the profiles Stein collects across those varying roads are camera image sequences, not inertial sensor-derived vehicle motion profiles of a vehicle portion. The amended claim 21 requires inertial sensor-based profiles, which Stein does not disclose regardless of the road conditions under which its training data is collected.” Filed 9/11/2025 has been fully considered but is not persuasive.
Applicant contends that Stein fails to disclose obtaining first vehicle motion profiles associated with road features and second vehicle motion profiles associated with an absence of road features, and further asserts that Stein’s training data consists only of camera image sequences rather than inertial based vehicle motion profiles.
As discussed above with respect to claim 1, Stein expressly teaches that vehicle motion measurements may be provided by inertial sensors integrated into the camera or mounted elsewhere on or in the vehicle. Further Stein teaches that the dataset supplied to the artificial neural network includes sensor movement information in addition to image sequences.
With respect to the first and second vehicle motion profiles, Stein teaches training the model using data collected from vehicles traversing numerous roads under varying geographic locations, road conditions, lighting conditions, and weather conditions. Such training data necessarily includes road segments exhibiting road surface features as well as road segments lacking such features. Therefore, Stein discloses obtaining vehicle motion information associated with both road segments containing road features and road segments lacking such road features.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure. WO 2018204656 A1 teaches detection and classification systems and methods for autonomous vehicle navigation.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Joshua J Penko whose telephone number is (571)272-2604. The examiner can normally be reached Monday thru Friday 8-5 ET.
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, Hitesh Patel can be reached at 571-270-5442. 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.
/JOSHUA JEFFREY PENKO/ Examiner, Art Unit 3667
/Hitesh Patel/ Supervisory Patent Examiner, Art Unit 3667
6/29/26
Prosecution Insights