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 .
Response to Appeal brief
Applicant’s arguments, see Appeal brief filed on 04/08/2026, with respect to the prior art 102 rejection(s) been fully considered and are persuasive. The examiner hereby withdraws the prior art rejections set forth in the previous Office Action mailed on 12/09/2025. However, upon further consideration, a new ground(s) of rejection is made, therefore, all of applicant’s arguments are moot in light of the new grounds of rejection.
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) 1-2, 5-6, 18, 20-27, 30 and 33 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kamo et al. (US 20160097853 A1) in view of Agrawal et al. (US 20190095756 A1).
Regarding claim 1.
Kamo teaches a method, comprising:
- receiving, from a radar, a first reflected signal (see ¶ 91, “During learning of the RBF neural network, the pattern that is fed to the input layer x.sub.i of the RBF neural network is the vector b expressed by Equation 6, which is obtained from the array reception signal X. On the other hand, the training signal that is fed is a signal (in vector expression) specifying the number of preceding vehicles (one or more) and their spatial distribution as existing when that vector b was obtained.”) and a second reflected signal (see ¶ 141, “The signal processing circuit 30 directly or indirectly receives reception signals from the array antenna AA, and inputs the reception signals, or a secondary signal (s) which has been generated from the reception signals, to the neural network NN. A part or a whole of the circuit (not shown) which generates a secondary signal(s) from the reception signals does not need to be provided inside of the signal processing circuit 30. A part or a whole of such a circuit (preprocessing circuit) may be provided between the array antenna AA and the radar signal processing apparatus 300.”);
- determining a reference signal of the first reflected signal (see ¶ 91, “the training signal that is fed is a signal (in vector expression) specifying the number of preceding vehicles (one or more) and their spatial distribution as existing when that vector b was obtained. This mapping will be repeatedly learned to a point where a signal which is output in response to any given pattern that is fed to the input layer x.sub.i will be a signal reflecting the learning results, i.e., one that accurately identifies the number of preceding vehicles (one or more) and their spatial distribution.”)
and training an artificial neural network using the first reflected signal and the reference signal of the first reflected signal (see ¶ 91, “the training signal that is fed is a signal (in vector expression) specifying the number of preceding vehicles (one or more) and their spatial distribution as existing when that vector b was obtained. This mapping will be repeatedly learned to a point where a signal which is output in response to any given pattern that is fed to the input layer x.sub.i will be a signal reflecting the learning results, i.e., one that accurately identifies the number of preceding vehicles (one or more) and their spatial distribution.”);
- upon training, determining an output of the artificial neural network associated with the first reflected signal (see ¶ 91, “the training signal that is fed is a signal (in vector expression) specifying the number of preceding vehicles (one or more) and their spatial distribution as existing when that vector b was obtained. This mapping will be repeatedly learned to a point where a signal which is output in response to any given pattern that is fed to the input layer x.sub.i will be a signal reflecting the learning results, i.e., one that accurately identifies the number of preceding vehicles (one or more) and their spatial distribution.”); and
- providing a magnitude and angle image of the radar associated with the second reflected signal based on the output of the artificial neural network associated with the first reflected signal (see ¶ 144, “The arriving wave estimation circuit AU is configured or programmed to estimate an angle representing the azimuth of each arriving wave by an arbitrary algorithm for direction-of-arrival estimation, and output a signal indicating the estimation result. The signal processing circuit 30 may be configured or programmed to estimate the distance to each target as a wave source of an arriving wave, the relative velocity of the target, and the azimuth of the target by using a known algorithm which is executed by the arriving wave estimation circuit AU, and output a signal indicating the estimation result.”, also see ¶ 150, “the signal processing circuit 30 may be configured or programmed to operate while switching between a first mode of utilizing the neural network NN, and a second mode of utilizing the arriving wave estimation circuit AU. There may be various conditions for travel control that stipulate switching between the first mode and the second mode. For example, the signal processing circuit 30 may select the first mode while ACC is activated, and the second mode while ACC is not activated.” and ¶ 151, “Once a spatial distribution of preceding vehicles is determined by the neural network NN, the number of preceding vehicles, i.e., the number of arriving waves, is able to be determined. So long as the number of arriving waves is known, eigenvalue decomposition for running a known algorithm for direction-of-arrival estimation becomes unnecessary. Stated otherwise, based on information of the number of arriving waves as detected by the neural network NN, it becomes possible to run an algorithm for direction-of-arrival estimation (e.g., the SAGE method) with a smaller computation amount than conventionally, and yet estimate the directions of preceding vehicles with a high accuracy.”),
- see ¶ 144, “FIG. 10, separately from the neural network NN, an arriving wave estimation circuit AU is provided in the signal processing circuit 30. The arriving wave estimation circuit AU is configured or programmed to estimate an angle representing the azimuth of each arriving wave by an arbitrary algorithm for direction-of-arrival estimation, and output a signal indicating the estimation result. The signal processing circuit 30 may be configured or programmed to estimate the distance to each target as a wave source of an arriving wave, the relative velocity of the target, and the azimuth of the target by using a known algorithm which is executed by the arriving wave estimation circuit AU, and output a signal indicating the estimation result. By providing such an arriving wave estimation circuit AU, it becomes possible to acquire position information of a preceding vehicle even in a situation where the neural network NN is unable to detect a spatial distribution pattern of vehicles, and utilize it for travel assistance.”, also see ¶ 151, teaches NN output (number of arriving waves) enables the system to run SAGE algorithm also see ¶¶ 238-247, teaches SAGE/EM algorithms computing angle/direction using figure 19 steps).
Kamo do not specifically teach upon training, determining an output of the artificial neural network associated with the first reflected signal; wherein the output of the artificial neural network comprises at least one algorithm selected by the artificial neural network, the method further comprising: determining the magnitude and angle image of the radar associated with the second reflected signal based on the at least one algorithm selected by the artificial neural network.
Agrawal teaches upon training, determining an output of the artificial neural network associated with the first reflected signal (see ¶ 93, “meta-models 151-153 may each be an already trained neural network that takes a subset of hyperparameter values and a subset of meta-feature values as stimulus inputs, shown as dashed arrows entering meta-models 151-153. ”, also see ¶ 95, ”Meta-models 151-153 are already trained regressors that process inputs to emit a comparative suitability score. For example, meta-model 151 emits score 161.”); wherein the output of the artificial neural network comprises at least one algorithm selected by the artificial neural network (see ¶¶ 95-98, “Meta-models 151-153 are already trained regressors that process inputs to emit a comparative suitability score. For example, meta-model 151 emits score 161. Scores 161-163 share a performance measurement scale…a score may simply be a comparative measure of abstract suitability. Regardless of score semantics, each meta-model of each algorithm emits a score. Computer 100 may select the best one or few algorithms (perhaps also best hyperparameter values), such as 122 as shown, as ranked based on sorted scores.”, also see ¶ 64-65, “One or more of the algorithms are selected based on the respective scores”), the method further comprising: determining the magnitude and angle image of the radar associated with the second reflected signal based on the at least one algorithm selected by the artificial neural network (see ¶ 99, “Computer 100 (or a downstream computer) may then use selected algorithm 122 to achieve a result, such as 190. For example, computer 100 may use inference dataset 110 (or a larger dataset that includes 110) to actually train one or a few alternate configurations of algorithm 122.”, also see ¶ 108-109, “After step 204 is sufficiently repeated, all meta-models of all algorithms 121-123 have scores. Based on those scores, at least one promising algorithm is selected for training. For example, computer 100 selects algorithm 122 that has the highest scoring meta-model of all algorithms or the highest mean, median, or modal score of all algorithms. Based on the inference dataset, step 208 invoked the selected algorithm(s) to obtain a result. This may or may not entail training at least one model (distinctly configured instance) of the selected algorithm.”).
Both Kamo and Agrawal pertain to the problem of neural network systems, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Kamo and Agrawal to teach the above limitations. The motivation for doing so would be “Decades of research have created a huge assortment of algorithms and techniques that can be applied to these applications. Selecting the best algorithm for an application may be difficult and resource intensive. For example, a classification task can be done by several algorithms such as support vector machines (SVMs), random forests, decision trees, artificial neural networks, and more. Each of these algorithms has many variations and configurations and performs differently for different datasets. Choosing the best algorithm is typically a manual task performed by a data scientist or a machine learning expert having years of experience.” (see Agrawal ¶ 2).
Regarding claim 2.
Kamo and Agrawal teaches the method according to claim 1,
Kamo further teaches wherein the method is for an apparatus comprising the artificial neural network (see ¶ 83, “FIG. 3 shows a structural model of a generic hierarchical neural network. Although the present specification will illustrate an RBF (radial basis function) neural network, which is a kind of hierarchical neural network, any other hierarchical neural network or non-hierarchical neural network may instead be used.” and figure 3).
Regarding claim 5.
Kamo and Agrawal teaches the method according to claim 1,
Kamo further teaches further comprising: - receiving the reference signal from a three-dimensional sensor, the three-dimensional sensor being associated with the radar (see ¶ 120 teaches a camera to recognize the patterns, also see ¶ 203, “the image processing circuit 52 is configured or programmed to estimate distance information of an object by detecting the depth value of an object within an acquired video, or detect size information and the like of an object from characteristic amounts in the video, thus detecting position information of the object.”, i.e. depth estimation and position information implies 3-dimensional)
Regarding claim 6.
Kamo and Agrawal teaches the method according to claim 1,
Kamo further teaches further comprising: - determining the reference signal from a simulated reference signal (see ¶ 30 and ¶ 31 specifying that show scenarios associated with training signal therefore, simulated scenes, also see ¶ 91 training signals).
Claim 18 recites a computer program product, embodied on a non-transitory computer readable medium, configured to control a processing unit to perform the method recited in claim 1. Therefore the rejection of claim 1 above applies equally here. Kamo also teaches the addition elements of claim 18 not recited in claim 1 comprising a non-transitory computer readable medium, configured to control a processing unit (see ¶ 23, “These general and specific aspects may be implemented using a system, a method, and a computer program stored on a computer readable medium, and any combination of systems, methods, and computer programs stored on a computer readable medium.”).
Claim 20 recites an apparatus comprising at least one processing unit, and at least one memory including computer program code to perform the method recited in claim 1. Therefore the rejection of claim 1 above applies equally here. Kamo also teaches the addition elements of claim 20 not recited in claim 1 comprising an apparatus comprising at least one processing unit, and at least one memory including computer program code (see figure 16, processor and memory, also see ¶ 23, “These general and specific aspects may be implemented using a system, a method, and a computer program stored on a computer readable medium, and any combination of systems, methods, and computer programs stored on a computer readable medium.”).
Claims 21-25 recites an apparatus comprising at least one processing unit, and at least one memory including computer program code to perform the method recited in claims 2-6. Therefore the rejection of claims 2-6 above applies equally here.
Regarding claim 26.
Kamo and Agrawal teaches the apparatus according to claim 20,
Kamo further teaches wherein the at least one memory and the computer program code are further configured to, with the at least one processing unit, cause the apparatus at least to: - determine a difference between an output of the artificial network associated with the first reflected signal and the reference signal; and - train the neural network using the difference (see ¶ 113, “At step S17, the signal processing circuit performs computation (forward computation) by using Equation 7 and Equation 8.”, also see ¶ 114, “At step S18, the signal processing circuit determines a mean squared error between the obtained results and the training signal serving as a reference, and performs computation (backward computation) of correcting the weights so as to minimize the mean squared error. The corrected weights are to be utilized in any subsequent learning.”).
Regarding claim 27.
Kamo and Agrawal teaches the apparatus according to claim 20,
Kamo further teaches wherein the at least one memory and the computer program code are further configured to, with the at least one processing unit, cause the apparatus at least to: - train the artificial neural network by providing the first reflected signal to an input layer of the artificial neural network; and - determine the output of the artificial network associated with the first reflected signal from an output layer of the artificial neural network (see ¶ 113, “At step S17, the signal processing circuit performs computation (forward computation) by using Equation 7 and Equation 8.”, also see ¶ 114, “At step S18, the signal processing circuit determines a mean squared error between the obtained results and the training signal serving as a reference, and performs computation (backward computation) of correcting the weights so as to minimize the mean squared error. The corrected weights are to be utilized in any subsequent learning.”).
Regarding claim 30.
Kamo and Agrawal teaches the apparatus according to claim 20,
Kamo further teaches wherein the output of the artificial neural network comprises at least one parameter selected by the artificial neural network for the at least one algorithm, and wherein the at least one memory and the computer program code are further configured to, with the at least one processing unit, cause the apparatus at least to:
- determine the magnitude and angle image of the radar associated with the second reflected signal based on the at least one parameter (see ¶ 151, “Once a spatial distribution of preceding vehicles is determined by the neural network NN, the number of preceding vehicles, i.e., the number of arriving waves, is able to be determined. So long as the number of arriving waves is known, eigenvalue decomposition for running a known algorithm for direction-of-arrival estimation becomes unnecessary. Stated otherwise, based on information of the number of arriving waves as detected by the neural network NN, it becomes possible to run an algorithm for direction-of-arrival estimation (e.g., the SAGE method) with a smaller computation amount than conventionally, and yet estimate the directions of preceding vehicles with a high accuracy.”, i.e. SAGE algorithm).
Claim 33 recites a method to perform the method similarly with small variation as recited in claim 1. Therefore the rejection of claim 1 above applies equally here.
Claim(s) 3-4 and 31-32 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kamo et al. (US 20160097853 A1) in view of Agrawal et al. (US 20190095756 A1) in view of Harrison et al. (US 20190391235 A1).
Regarding claim 3.
Kamo and Agrawal teaches the method according to claim 1,
Kamo and Agrawal do not specifically teach the limitations of claim 3.
Harrison teaches further comprising: - transmitting the second reflected signal as an input to the artificial neural network; and - receiving the output of the artificial neural network, wherein the output comprises the magnitude and angle image of the radar associated with the second reflected signal (see figure 7 and ¶ 33 “The radar data is first sent to the super-resolution network 706, which is, or includes at least a portion of, the super-resolution network 200, 300 and/or 400, to increase the resolution of the input radar set to generate a higher resolution radar dataset. In some aspects, the high resolution radar dataset may include features of the input radar set along with features that substantially correspond to a target lidar dataset. The higher resolution radar data is then sent to a target identification and decision module 708, which implements a convolutional neural network for target detection and identification and a decision neural network for deciding which actions the antenna module 702 should perform next. For example, the target identification and decision module 708 may detect a cyclist on the path of the ego vehicle and may direct the antenna module 702 to focus additional RF beams at a given phase shift and direction within the portion of the FoV corresponding to the cyclist's location.”).
Kamo, Agrawal and Harrison pertain to the problem of neural network radar systems, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Kamo, Agrawal and Harrison to teach the above limitations. The motivation for doing so would be “The system includes a super-resolution network configured to receive the radar data with the first resolution and produce radar data with a second resolution different from the first resolution using first neural networks. The system also includes a target identification module configured to receive the radar data with the second resolution and to identify the detected target from the radar data with the second resolution using second neural networks. Other examples disclosed herein include a method of operating the radar system in the autonomous driving system of the ego vehicle.” (See Harrison e.g. ¶ Abstract) and “increasingly assume control of driving functions such as steering, accelerating, braking and monitoring the surrounding environment and driving conditions to respond to events, such as changing lanes or speed when needed to avoid traffic, crossing pedestrians, animals, and so on.” (see Harrison ¶ 2).
Regarding claim 4.
Kamo, Agrawal and Harrison teach the method according to claim 3,
Harrison teaches wherein the magnitude and angle image of the radar associated with the second reflected signal is determined by the artificial neural network using an algorithm modelled by the artificial neural network during said training (see figure 5 element 512 and ¶ 28 “at step 512, the super-resolution network is then trained with the training set to map the radar data into the lidar data. For example, the super-resolution network can learn to map features of a coarse-resolution radar image into an output image having features that substantially correspond to features of a target fine-resolution lidar image. The neural network performance can improve even further with a larger training dataset.”. )
The motivation utilized in the combination of claim 3, super, applies equally as well to claim 4.
Regarding claim 31.
Kamo and Agrawal teaches the apparatus according to claim 20,
Kamo and Agrawal do not specifically teach the limitations of claim 31.
Harrison teaches wherein an output of a first algorithm associated with the second reflected signal is weighted by a first weight and an output of a second algorithm is weighted by a second weight and the magnitude and angle image of the radar is determined by combining the weighted output of the first algorithm and the weighted output of the second algorithm (see ¶ 36, “The RF beams reflect from targets in the ego vehicle's path and surrounding environment, and the RF reflections are received by the transceiver module 808. Radar data from the received RF beams is provided to the perception module 804 for target detection and identification. A super-resolution network 812 increases the resolution of the radar data prior to it being processed to detect and identify targets. For example, the super-resolution network 812 can process the radar data and determine high resolution radar data for use by the perception module 804. In various examples, the super-resolution network 812 can be a part of the perception module 804, such as on the same circuit board as the other modules within the perception module 804. Also, in various examples, the data encoding may use the lidar point cloud from the ego lidar to perform NLOS correction in the radar data.”, i.e. to use super resolution network for radar as a first algorithm and additionally uses the lidar point cloud as a second algorithm. The combination of both exhibits inherently two weights).
The motivation utilized in the combination of claim 3, super, applies equally as well to claim 31.
Regarding claim 32.
Kamo, Agrawal and Harrison teach the apparatus according to claim 31,
Harrison further teaches wherein the first weight and the second weight are modelled by the artificial neural network (see ¶ 36 and ¶ 33, “The higher resolution radar data is then sent to a target identification and decision module 708, which implements a convolutional neural network for target detection and identification and a decision neural network for deciding which actions the antenna module 702 should perform next.”).
The motivation utilized in the combination of claim 3, super, applies equally as well to claim 32.
Claim(s) 29 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kamo et al. (US 20160097853 A1) in view of Agrawal et al. (US 20190095756 A1) in view of Liu et al. (US 11468318 B2).
Regarding claim 29.
Kamo and Agrawal teaches the apparatus according to claim 20,
Kamo and Agrawal do not teach the limitation in claim 29.
Liu teaches wherein the at least one algorithm comprises a mirroring algorithm, the Burg algorithm, interpolation and/or extrapolation (see col 4 lines 65-68, “Frame interpolation for video is one of the basic computer vision and video processing technologies. It is a special case of image-based rendering where middle frames are interpolated from temporally neighboring frames. This section focuses on research that is specific to video frame interpolation.”).
Kamo, Agrawal and Liu pertain to the problem of neural network radar systems, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Kamo, Agrawal and Liu to teach the above limitations. The motivation for doing so would be “Video frame interpolation is a basic video processing technique that is used to generate intermediate frames between any two consecutive original frames. Video frame interpolation algorithms typically estimate optical flow or its variations and use them to warp and blend original frames to produce interpolation results.” (see Liu col 1 lines 34-40).
Related art:
Stachnik et al. (US 11448746 B2) teaches estimating a velocity magnitude of a moving target in a horizontal plane using radar signals received by a radar detection system, the radar detection system being configured to resolve multiple dominant points of reflection, i.e. to receive a plurality of radar signals from the moving target in a single measurement instance of a single, wherein each of the resolved points of reflection is described by data relating to a range, an azimuth angle and a raw range rate of the points of reflection in said single radar measurement instance. The invention further relates to a radar detection system.
Jonas et al. (US 20190056477 A1) teaches radar system includes a first sensor that generates a first chirp signal; a second sensor for generating a second chirp signal and which received reflected signals. One of the first sensor and second sensor receives a signal that includes a first reflected signal related to the first chirp signal and a second reflected signal related to the second chirp signal. A processor multiplies the received signal by one of the first chirp signal and the second chirp signal to obtain a desired signal indicative of one of the first reflected signal and the second reflected signal and an interference signal indicative of the other of the first reflected signal and the second reflected signal, and applies a filter to the mixed signal to separate the interference signal from the desired signal.
SAVCHENKOV et al. (US 20200256979 A1) teaches a signal source providing a transmission signal; a first transmitter for transmitting the signal as a first FMCW to a target; a second transmitter for transmitting the signal as a second FMCW to the target, where the first and second transmitters are spaced apart by a distance greater than a wavelength of the transmission signal; a receiver for receiving a reflected signal from the target; and an analyzer for determining the target distance and angle of the target with respect to the transmitters based on the received signal and the transmission signal. Radar devices with multiple input, multiple output transmitters are also described.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IMAD M KASSIM whose telephone number is (571)272-2958. The examiner can normally be reached 10:30AM-5:30PM, M-F (E.S.T.).
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, Michael J. Huntley can be reached at (303) 297 - 4307. 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.
/IMAD KASSIM/Primary Examiner, Art Unit 2129
Prosecution Insights