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 .
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) 1-7 and 10-22 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roajanasiri, et al. (“Roajanasiri”) (U.S. Pub. 2010/0108256) in view of Cybulsky, et al. (“Cybulsky”) (U.S. Pub. 2019/0384329).
Regarding claims 1 and 21, Roajanasiri discloses a system for fluid dispensing process control of a dispensing device having a nozzle (210) for dispensing a portion of viscous fluid (“adhesive dots”) according to a first value of an operating parameter (¶ [0011]: “a reference characteristic value”) of the dispensing device, the system comprising:
a controller (214) configured to generate one or more signals to: determine, based on data from a sensor (monitoring device 206 and/or weight measurements along conveyor belt (¶ [0017])), a characteristic (¶ [0016]: diameter, height, weight or location) of the portion of viscous fluid;
input the characteristic of the portion of viscous fluid and the first value of the operating parameter to a feedback control system (¶ [0019]: “The feedback system 208 may comprise any of a variety of systems communicatively coupled to and configured to receive information from the adhesive monitoring device 206. The feedback system 208 may be further configured to analyze such information, and modify a control variable associated with the adhesive dispenser 202”); and
determine, using the feedback control system, a defect classification (the resultant of the comparison of the average to a function of a reference characteristic value and the statistical dispersion) of the portion of viscous fluid based on the characteristic of the portion of viscous fluid and the first value of the operating parameter. (¶ [0022]: “determine an average for the characteristics of the plurality of adhesive dots 203; determine a statistical dispersion for the characteristics of the plurality of adhesive dots 203; compare the average to a function of a reference characteristic value and the statistical dispersion”.)
Roajanasiri particularly teaches comparing measured characteristics of the dispensed adhesive dots with a reference characteristic and modifying a dispensing control variable based on the comparison. (¶ [0022]: “modify a control variable associated with the adhesive dispenser 202 based at least in part on the comparison.”)
Roajanasiri determines the modification of the dispensing control parameter using its statistical/reference comparison (Abstract) rather than a machine-learning tool configured to receive the measured characteristic and operating parameter and determine a defect classification.
Cybulsky teaches a machine-learning process-control system in which process-control parameters and sensor/image data are provided to a machine-learning module, which determines a relationship between the process-control parameters and process outputs. Cybulsky further teaches using the machine-learned relationship to determine control-parameter values for controlling the process. (see Cybulsky: claim 1) Cybulsky further discloses machine-learning classification techniques, including support-vector machines, K-nearest neighbors, regression analysis and artificial neural networks. (see Cybulsky: claim 10)
Therefore, it would have been obvious to one with ordinary skill in the art, prior to the effective filing date of the claimed invention, to substitute the statistical/reference-comparison analysis of Roajanasiri with the machine-learning analysis of Cybulsky, thereby using a machine-learning tool to analyze the measured characteristic of the dispensed adhesive and the corresponding dispensing operating parameter and determine a defect classification of the dispensed adhesive, since doing so would be a mere substitution of one known analysis system for another analysis system with the expected results that the substituted analysis system would determine a defect classification based on the characteristic of the portion of viscous fluid and the first value of the operating parameter. (see MPEP 2143 I B)
Regarding claim 2, Roajanasiri’s controller is further configured to generate a signal to determines a change to a physical parameter (second value) of a subsequent adhesive dot based on measured characteristics (defect classification) of previously dispensed adhesive dots. (see Roajanasiri: claim 1) It specifically teaches modifying the pressure applied to the adhesive dispenser or the dispensing time interval to control the subsequent adhesive dot. (see Roajanasiri: claims 15 and 16)
Cybulsky teaches determining respective values of process-control parameters based on a relationship determined by machine learning and controlling the process using those values. (see Cybulsky: claim 1)
Thus, the combination determines a second operating-parameter value based on the machine-learned defect classification.
Regarding claim 3, the claim requires that the machine-learning tool employ a machine learning algorithm. The combination, as modified by Cybulsky, expressly teaches machine-learning algorithms. (see Cybulsky, algorithms listed in claim 10)
Regarding claim 4, the claim requires the machine-learning algorithm is selected from the group consisting of Deep Neural Network (DNN), eXtreme Gradient Boosting (XGBoost), Convolutional Machine Learning (CNN), Support Vector Machine (SVM), Multiple Linear Regression, Random Forest, AdaBoost, Artificial Neural Network Tool (ANN), Decision Tree (DT), Naive Bayes, K Nearest Neighbor (KNN) and other listed algorithms.
The combination, as modified by Cybulsky, expressly teaches SVM, K-nearest neighbors, regression analysis, and artificial neural networks. (see Cybulsky: claim 10)
The particular selection of a known machine-learning algorithm for performing the machine-learning classification would have been an obvious matter of selecting from known alternatives according to the particular characteristics of the dispensing-control problem.
Regarding claim 5, the claim requires the machine learning tool utilizes one or more feature vector processes, classification processes, grouping processes, regression processes, analysis processes, matching processes, training processes, or diagnostic processes.
Cybulsky expressly teaches classification and related machine-learning analysis. For example, claim 10 of Cybulsky identifies algorithms including LDA, KNN, SVM, regression analysis and ANN. Cybulsky further discloses training KNN and SVM models using process data to distinguish different process conditions. (Cybulsky: ¶ [0110])
Regarding claim 6, the claim requires an in-training machine learning tool.
Cybulsky expressly teaches training machine-learning algorithms using sensor data to classify different process conditions. In paragraph [0110] of Cybulsky, Example 3, LDA, KNN, and SVM were trained using acoustic process data and subsequently validated against excluded data.
Regarding claim 7, requires the defect classification of the portion of viscous fluid is determined by the machine learning tool based on a training of the machine learning tool to associate images of dispensed portions, and operating parameters associated with dispensed portions, with defect classifications.
Cybulsky teaches using optical sensor image data together with control parameters and process outputs as inputs to a machine-learning module. (Cybulsky: claim 1)
Cybulsky further teaches training machine-learning system to distinguish different process conditions from sensor data. (Cybulsky: claim 1)
Roajanasiri discloses the dispenser and measured characteristics of adhesive dots, including machine-vision measurement of the dots.
Therefore, it would have been obvious to one with ordinary skill in the art, prior to the effective filing date of the claimed invention, to train the machine-learning system using images/characteristics of the dispensed adhesive portions together with the associated dispensing parameters and to associate those inputs with defect classifications.
Regarding claim 10, Roajanasiri teaches a machine-vision system for measuring the diameter of each adhesive dot. (claim 3) Cybulsky teaches an optical sensor configured to generate image data indicative of the process. (claim 1)
Regarding claim 11, Roajanasiri, as modified by Cybulsky, expressly teaches an image-data signal processing module configured to receive image data and transform the image-data signal into an image. (Cybulsky: claim 11) Image processing constitutes preprocessing of image data prior to machine-learning analysis.
Regarding claim 12, Roajanasiri teaches machine-vision measurement of dispensed adhesive dots. (claim 3) Cybulsky teaches optical monitoring and analysis of images of a fluid/plasma process, including image data obtained during the process (claim 1) and it would have been obvious to obtain the machine-vision data during dispensing, including while the dispensed material is in flight, to provide earlier feedback for the closed-sloop dispensing process.
Regarding claim 13, Roajanasiri teaches an adhesive monitoring device positioned downstream from the adhesive dispenser. (claim 18) Positioning the machine-vision camera downstream and below the dispensing nozzle to view the dispensed adhesive would have been an obvious matter of positional arrangement dictated by the desired field of view and the physical geometry of the dispensing operation.
Regarding claim 14, Cybulsky teaches optical sensing and image-data acquisition for analyzing process conditions. (Cybulsky: claim 1) Using multiple views of a dispensed adhesive portion to obtain additional or positional information would have been a known image-inspection technique and an obvious modification of the machine-vision system of the primary reference.
Regarding claim 15, Roajanasiri teaches machine-vision inspection of adhesive dots. (claims 3 and 20)
Cybulsky teaches optical monitoring of a fluid spray process and acquisition of images using optical sensing, including optical elements such as “light sources”. (¶ [0056]).
Providing illumination across the dispensing path to improve optical detection of the dispensed material would have been an obvious implementation of the optical inspection system.
Regarding claim 16, the combination of the optical inspection system of Roajanasiri and the optical sensing of the Cybulsky provides the claimed optical inspection arrangement and “light sources”. The use of a light beam intersecting the dispensing path to improve image contrast and detect the dispensed material would have been an obvious implementation choice.
Regarding claim 17, Roajanasiri teaches measuring a diameter, height or weight of each adhesive dot using the measured characteristic for closed-loop control. (¶ [0016]).
Regarding claim 18, Roajanasiri discloses a scale. (¶ [0017])
Regarding claim 19, Roajanasiri discloses the characteristic of the portion of viscous fluid comprises a directionality of the portion of viscous fluid. (¶ [0028]: “location characteristics (e.g., absolute or relative positioning information)”)
Regarding claim 20, Roajanasiri discloses the characteristic of the portion of viscous fluid comprises size characteristics (¶ [0028]: “size characteristics (e.g., a diameter, height, weight, and/or circumference)”) with which a volume could easily have been calculated.
Regarding claims 22, Roajanasiri discloses a system capable of fluid dispensing process control of a dispensing device having a nozzle (210) for dispensing a portion of viscous fluid (“adhesive dots”) according to a first value of an operating parameter (¶ [0011]: “a reference characteristic value”) of the dispensing device, the system comprising:
a controller (214) configured to generate one or more signals to: determine, based on data from a sensor (monitoring device 206 and/or weight measurements along conveyor belt (¶ [0017])), a characteristic (¶ [0016]: diameter, height, weight or location) of the portion of viscous fluid;
inputting the characteristic of the portion of viscous fluid and the first value of the operating parameter to a feedback control system (¶ [0019]: “The feedback system 208 may comprise any of a variety of systems communicatively coupled to and configured to receive information from the adhesive monitoring device 206. The feedback system 208 may be further configured to analyze such information, and modify a control variable associated with the adhesive dispenser 202”); and
determining, using the feedback control system, a defect classification (the resultant of the comparison of the average to a function of a reference characteristic value and the statistical dispersion) of the portion of viscous fluid based on the characteristic of the portion of viscous fluid and the first value of the operating parameter. (¶ [0022]: “determine an average for the characteristics of the plurality of adhesive dots 203; determine a statistical dispersion for the characteristics of the plurality of adhesive dots 203; compare the average to a function of a reference characteristic value and the statistical dispersion”.)
Roajanasiri particularly teaches comparing measured characteristics of the dispensed adhesive dots with a reference characteristic and modifying a dispensing control variable based on the comparison. (¶ [0022]: “modify a control variable associated with the adhesive dispenser 202 based at least in part on the comparison.”)
Roajanasiri determines the modification of the dispensing control parameter using its statistical/reference comparison (Abstract) rather than a machine-learning tool configured to receive the measured characteristic and operating parameter and determine a defect classification.
Cybulsky teaches a machine-learning process-control system in which process-control parameters and sensor/image data are provided to a machine-learning module, which determines a relationship between the process-control parameters and process outputs. Cybulsky further teaches using the machine-learned relationship to determine control-parameter values for controlling the process. (see Cybulsky: claim 1) Cybulsky further discloses machine-learning classification techniques, including support-vector machines, K-nearest neighbors, regression analysis and artificial neural networks. (see Cybulsky: claim 10)
Therefore, it would have been obvious to one with ordinary skill in the art, prior to the effective filing date of the claimed invention, to substitute the statistical/reference-comparison analysis of Roajanasiri with the machine-learning analysis of Cybulsky, thereby using a machine-learning tool to analyze the measured characteristic of the dispensed adhesive and the corresponding dispensing operating parameter and determine a defect classification of the dispensed adhesive, since doing so would be a mere substitution of one known analysis system for another analysis system with the expected results that the substituted analysis system would determine a defect classification based on the characteristic of the portion of viscous fluid and the first value of the operating parameter. (see MPEP 2143 I B)
Claim(s) 8 and 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over Roajanasiri and Cybulsky as applied to claim 1 above, and further in view of Zhang, et al. (“Zhang”) (U.S. Pub. 2019/0370955).
Regarding claim 8, the combination discloses determining a defect classification of a dispensed fluid portion using a machine-learning tool based on the characteristic of the portion and an operating parameter.
Zhang teaches a defect classifier for classifying defects detected from images and determining probability and uncertainty associated with defect classification. (Zhang: claims 1, 18 and 19) Zhang further teaches selecting data based on the probability or uncertainty that the data represents a particular defect type. (Claims 28-29)
Therefore, it would have been obvious to one with ordinary skill in the art, prior to the effective filing date of the claimed invention, to apply the probabilistic classification of Zhang to the machine-learning defect classification of the combination to provide an indication of the probability that a dispensed fluid portion belongs to the determined defect classification. Such a modification would be simple substitution of one known machine-learning classification output for another to obtain the predictable results of providing a probability associated with the classification.
Regarding claim 9, Roajanasiri discloses monitoring successive dispensed adhesive portions and analyzing their characteristics to detect changes in dispensing performance and modifying the dispensing control parameter. (¶ [0057]-[0064])
Zhang teaches determining probability and uncertainty associated with defect classification (claims 18-19) and it would have been obvious to one with ordinary skill in the art, prior to the effective filing date of the claimed invention, to monitor the probability associated with successive defect classifications and determine a trend in probabilities and to use such a trend as an indication of degradation of the dispensing device (¶ [0106]: “identify all failure types”) and to predict a time of component failure, thereby providing the predictable benefit of identifying impending failure before unacceptable dispensing occurs.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. See form PTO-892, attached.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL J MELARAGNO whose telephone number is (571)270-7735. The examiner can normally be reached Mon - Fri: 8 am - 5 pm +/- flex.
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, Paul Durand can be reached at (571) 272-4459. 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.
/MICHAEL J. MELARAGNO/ Examiner, Art Unit 3754
Prosecution Insights