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 .
Status of Claims
This action is in reply to the amendments and remarks filed on 06/24/2025.
Claims 1-2, 4-5, 7-11, 13-14, 16-19, 21-23, and 25-33 are pending.
Claims 1-2, 4-5, 13, 21, 23, and 25 have been amended.
Claims 3, 6, 12, 15, 20, and 24 have been canceled.
Claims 27-33 have been added.
Response to Arguments
Applicant’s arguments, with respect to the rejection(s) of claim(s) 1 and 13 under 35 U.S.C. 103, have been considered but they are not persuasive. Applicant argues that no art of reference teaches the amended claim language of claims 1 and 13 that now state “predicting, by the device, additional sensor data values of the sensor using the set of machine learning model parameters with the at least one sensor data stream and the predictive machine learning model configured by the set of machine-learned values”, since “Liang does not disclose predicting a directional velocity as indicated in sensor data, but rather controlling the robot to achieve a directional velocity [and]… that predicting values of a sensor data stream of a sensor is not equivalent to generating control instructions that may alter values of the sensor data stream of the sensor (including a camera or LIDAR sensor) in the future”. The examiner respectfully disagrees.
It is noted that the argued “predicting a directional velocity as indicated in sensor data” or “generating control instructions that may alter values of the sensor data stream of the sensor…in the future” are not claimed in amended claim 1, and therefore the applicant is arguing a narrower scope then what is claimed.
Liang, sections 3.2-3.4 and 4.1 teach the trained algorithm (using the set of machine learning model parameters…and the predictive machine learning model configured by the set of machine-learned values) to determine best policy actions to use given a robot’s “multiple sensors” giving “sensor streams” (with the at least one sensor data stream) in order to control the robot’s directional “velocity” (predicting…additional sensor data values of the sensor). Section 5.1 teaches using a processor (by the device) and GPU for implementing the operations of the embodiments. Further, Liang teaches outputting robot trajectories (direction as argued) and velocities according to the learned policy.
See 35 U.S.C 103 section for full mapping of claim limitations necessitated by applicant amendments.
Applicant’s arguments, with respect to the rejection(s) of claim(s) 1 and 13 under 35 U.S.C. 103, have been considered but they are not persuasive. Applicant argues that no art of reference teaches the amended claim language of claims 1 and 13 that now state “receiving, by the device, at least one sensor data stream generated by a sensor”, since “sensor data streams that create the observation vector data of the robot, as disclosed in Liang, cannot be cited as both the ‘set of machine learning model parameters’ and the ‘sensor data stream’ recited in claim 1”. The examiner respectfully disagrees.
Specification, paragraphs 0052-0053 state “the set of model parameters comprise at least one vector. In some of these embodiments, the at least one vector comprises a motion vector associated with a robot.”
Liang, sections 3.2-3.4 and 4.1 teach an obstacle avoiding robot using “multiple sensors” giving “sensor streams” (receiving…at least one sensor data stream that is generated by a sensor) that create another “robot’s observation vector” data including the robot’s current velocity. Section 5.1 teaches using a processor (by the device) and GPU for implementing the operations of the embodiments. Further, sections 4.1 and 5.1-5.3 teach creating outputs for the “datasets” in training (predictions) and then applying the model to evaluation test sets of sensor data (additional sensor data); thus, creating another vector as mapped above.
See 35 U.S.C 103 section for full mapping of claim limitations necessitated by applicant amendments.
Applicant’s arguments, with respect to the rejection(s) of claim(s) 1 and 13 under 35 U.S.C. 103, have been considered but they are not persuasive. Applicant argues that no art of reference teaches the amended claim language of claims 1 and 13 that now states “receiving, by a device and from a transmitting device, a set of machine learning model parameters associated with a predictive machine learning model, wherein the set of machine learning model parameters received by the device includes a set of machine-learned values that configures the predictive machine learning model to predict data values of sensors using at least one sensor data stream generated by the sensors”, since canceled claim 24 recited “the set of machine learning model parameters includes neural network weights and biases” and Lajevardi fails to teach anything more than a generic teaching of transmitting parameters. The examiner respectfully disagrees.
It is noted that the argued canceled limitation is not claimed in amended claim 1, and therefore the applicant is arguing a narrower scope then what is claimed. The argued limitation is now recited in dependent claim 27.
Specification, paragraphs 0052-0053 state “the set of model parameters comprise at least one vector. In some of these embodiments, the at least one vector comprises a motion vector associated with a robot.”
Liang, sections 3.2-3.4 and 4.1 teach the training (configures) neural network “navigation algorithm” (predictive machine learning model) with the “the robot’s observation vector” data including the robot’s current velocity (the set of machine learning model parameters…includes a set of machine-learned values) to determine best policy actions to use given the sensor vector data from “sensors” (using at least one sensor data stream generated by the sensors) in order to control the robot’s directional “velocity”. Section 5.1 teaches using a processor and GPU for implementing the operations of the embodiments.
See 35 U.S.C 103 section for full mapping of claim limitations necessitated by applicant amendments.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-2, 4-5, 7-8, 10-11, 13-14, 16-19, 21-23, 25-26, 28-31 are rejected under 35 U.S.C. 103 as being unpatentable over Liang et al (“Realtime Collision Avoidance for Mobile Robots in Dense Crowds using Implicit Multi-sensor Fusion and Deep Reinforcement Learning”, 2020) hereinafter Liang, in view of Shao et al (“A Deep Learning Approach for Fault Diagnosis of Induction Motors in Manufacturing”, 2017) hereinafter Shao.
Regarding claims 1 and 13, Liang teaches a method for predicting data values of data sources, and information technology system for predicting data values of data sources, the system comprising: a predictive machine learning model device configured to (sections 3.2-3.4, 4.1, and 5.1 teach using a processor for implementing the operations of the embodiments including collecting robot “sensor streams” and using an algorithm for predicting the best actions for avoiding detected obstacles), and the method comprising:
receiving, by a device and from a transmitting device, a set of machine learning model parameters associated with a predictive machine learning model (Examiner note: specification, paragraphs 0052-0053 state “the set of model parameters comprise at least one vector. In some of these embodiments, the at least one vector comprises a motion vector associated with a robot.”
Liang, sections 3.2-3.4 and 4.1 teach an obstacle avoiding robot creating “the robot’s observation vector” data including the robot’s current velocity (receiving…a set of machine learning model parameters) that is used to train a neural network “navigation algorithm” (associated with a plurality of predictive machine learning model) by a “training module” (from a transmitting device). Section 5.1 teaches using a processor (by the device) and GPU for implementing the operations of the embodiments.), and
wherein the set of machine learning model parameters received by the device includes a set of machine-learned values that configures the predictive machine learning model to predict data values of sensors using at least one sensor data stream generated by the sensors (sections 3.2-3.4 and 4.1 teach the training (configures) neural network “navigation algorithm” (predictive machine learning model) with the “the robot’s observation vector” data including the robot’s current velocity (the set of machine learning model parameters…includes a set of machine-learned values) to determine best policy actions to use given the sensor vector data from “sensors” (using at least one sensor data stream generated by the sensors) in order to control the robot’s directional “velocity”. Section 5.1 teaches using a processor and GPU for implementing the operations of the embodiments.);
receiving, by the device, at least one sensor data stream generated by a sensor (sections 3.2-3.4 and 4.1 teach an obstacle avoiding robot using “multiple sensors” giving “sensor streams” (receiving…at least one sensor data stream that is generated by a sensor) that create another “robot’s observation vector” data including the robot’s current velocity. Section 5.1 teaches using a processor (by the device) and GPU for implementing the operations of the embodiments.); and
predicting, by the device, additional sensor data values of the sensor using the set of machine learning model parameters with the at least one sensor data stream and the predictive machine learning model configured by the set of machine-learned values (sections 3.2-3.4 and 4.1 teach the trained algorithm (using the set of machine learning model parameters…and the predictive machine learning model configured by the set of machine-learned values) to determine best policy actions to use given a robot’s “multiple sensors” giving “sensor streams” (with the at least one sensor data stream) in order to control the robot’s directional “velocity” (predicting…additional sensor data values of the sensor). Section 5.1 teaches using a processor (by the device) and GPU for implementing the operations of the embodiments.).
Liang at least implies predicting, by the device, additional sensor data values of the sensor using the set of machine learning model parameters with the at least one sensor data stream and the predictive machine learning model configured by the set of machine-learned values (see mappings above), however Shao teaches predicting, by the device, additional sensor data values of the sensor using the set of machine learning model parameters with the at least one sensor data stream and the predictive machine learning model configured by the set of machine-learned values (sections 1 and 4.1-4.2 teach training a DBN model on a observed vibration signal training set (the predictive machine learning model configured by the set of machine-learned values), and using the trained DBN model (using…the predictive machine learning model) with initialized “parameters” (using the set of machine learning model parameters) to classify motor vibration signals (alternative predicting…additional sensor data values of the sensor), that are “pre-processed” as features for input to the model (predicting…additional sensor data values of the sensor/with the at least one data stream), for determining the degree and type of a motor issue (predicting…additional sensor data values of the sensor) and effectively diagnose motor failures for better maintenance.).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to implement Shao’s teachings of diagnosing motor faults for maintenance into Liang’s teaching of determining robot movement instruction vectors from sensor streams for avoiding moving obstacles in order to “enhance reliability and reduce costs for operation and maintenance of the manufacturing equipment” (Shao, section 1).
Regarding claims 2 and 14, the combination of Liang and Shao teach all the claim limitations of claims 1 and 13 above; and further teach further comprising receiving, by the device and from the transmitting device, the set of machine learning model parameters to the device in response to a high priority being associated with the at least one sensor data stream (Liang, sections 3.2-3.4, 4.1, and 5.1 teach “data from the 2-D lidar and the camera are synchronized”, but if a moving obstacle is detected, the camera stream is primarily used for determining “approximate positions, orientations and velocities of all obstacles in the frame” (in response to a high priority) within the model (associated with the at least one sensor data stream) for avoiding the obstacle before returning to the optimal path; for the processor to train the model via the “training module” on the sensor vectors on the processor or GPU devices (receiving, by the device and from the transmitting device, the set of machine learning model parameters to the device)).
Regarding claim 4, the combination of Liang and Shao teach all the claim limitations of claim 3 above; and further teach further comprising receiving, by the device and from the transmitting device, the set of machine learning model parameters in response to the at least one sensor data stream or the set of machine learning model parameters being unusual (Liang, sections 3.2-3.4, 4.1, and 5.1 teach “data from the 2-D lidar and the camera are synchronized”, but if a moving obstacle is detected in the data sensor streams (in response to the at least one sensor data stream or the set of machine learning model parameters being unusual), the camera stream is primarily used for determining “approximate positions, orientations and velocities of all obstacles in the frame” for avoiding the obstacle before returning to the optimal path; for the processor to train the model via the “training module” on the sensor vectors on the processor or GPU devices (receiving, by the device and from the transmitting device, by the transmitting device, the set of machine learning model parameters to the device)).
Regarding claim 5, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach further comprising receiving, by the device and from the transmitting device, the set of machine learning model parameters in response to a value of the set of machine learning model parameters changing from a previous value of the set of machine learning model parameters (Liang, sections 3.2-3.4, 4.1, and 5.1 teach detecting a moving obstacle in the data sensor streams vectors (in response to a value of the set of machine learning model parameters changing from a previous value of the set of machine learning model parameters), the camera stream is primarily used for determining “approximate positions, orientations and velocities of all obstacles in the frame” as input vectors (parameters changing from a previous value) for avoiding the obstacle before returning to the optimal path; for the processor to train the model via the “training module” on the sensor vectors on the processor or GPU devices (receiving, by the device and from the transmitting device, by the transmitting device, the set of machine learning model parameters to the device)).
Regarding claims 7 and 16, the combination of Liang and Shao teach all the claim limitations of claims 1 and 13 above; and further teach wherein the additional sensor data values of the sensor include a motion vector associated with a robot (Liang, sections 3.2-3.4 and 4.1 teach an obstacle avoiding robot using “multiple sensors” giving “sensor streams” that create “the robot’s observation vector” data including the robot’s current velocity (additional sensor data values of the sensor include a motion vector associated with a robot) that is used to train a neural network “navigation algorithm”).
Regarding claims 8 and 17, the combination of Liang and Shao teach all the claim limitations of claims 7 and 16 above; and further teach wherein the additional sensor data values of the sensor include at least one predicted location of a robot (Liang, sections 3.2-3.4 and 4.1 teach the trained algorithm to determine best policy actions to use given then sensor vector data in order to control the robot’s directional “velocity” of where and how to move (additional sensor data values of the sensor include at least one predicted locations of the robot) to avoid the detected obstacle).
Regarding claims 10 and 18, the combination of Liang and Shao teach all the claim limitations of claims 1 and 13 above; and further teach wherein the predictive machine learning model is at least one of: a behavior analysis machine learning model, an augmentation machine learning model, or a classification machine learning model (Liang, sections 3-3.4 and 4.1-4.2 teach collecting “sensor streams” of data to detect “behaviors of dynamic agents and pedestrians”/obstacles using an algorithm (behavior analysis machine learning model) and the algorithm for predicting robot “Critical behaviors” for movement in avoiding obstacles while deterring “oscillatory behaviors” (behavior analysis machine learning model)).
Regarding claims 11 and 19, the combination of Liang and Shao teach all the claim limitations of claims 1 and 13 above; and further teach wherein the sensor includes at least one of a set of radio frequency identification (RFID) sensors, a set of security cameras, or a set of vibration sensors (Liang, sections 3.2-3.4, 4.1, and 6 teach “We present a novel sensor-based navigation algorithm, CrowdSteer, that simultaneously uses multiple sensors such as 2-D lidars and cameras (set of security cameras)” that allows a robot to safely avoid moving obstacles (security)).
Regarding claim 21, the combination of Liang and Shao teach all the claim limitations of claim 13 above; and further teach wherein: the device receives a plurality of sensor data streams from a plurality of sensors, each data stream of the plurality of sensor data streams is generated by a respective sensor of the plurality of sensors, and each data stream of the plurality of sensor data streams is associated with a corresponding predictive machine learning model (Liang, sections 3.2-3.4, 4.1, and 5.1 teach an obstacle avoiding robot using “multiple sensors” and a “processor” giving “sensor streams” (each data stream of the plurality of sensor data streams is generated by a respective sensor of the plurality of sensors) that create “the robot’s observation vector” data including the robot’s current velocity that is used to train a neural network “navigation algorithm” (each data stream of the plurality of sensor data streams is associated with a corresponding predictive machine learning model); for the processor to train the model via the “training module” on the obtained sensor vectors on the processor or GPU devices (device receives a plurality of sensor data streams from a plurality of sensors)).
Regarding claim 22, the combination of Liang and Shao teach all the claim limitations of claim 13 above; and further teach wherein: the sensor includes at least one security camera, and the at least one sensor data stream at least partially includes at least one motion vector extracted from video data captured by the at least one security camera (Liang, sections 3.2-3.4, 4.1, and 6 teach “We present a novel sensor-based navigation algorithm, CrowdSteer, that simultaneously uses multiple sensors such as 2-D lidars and cameras (sensor includes at least one security camera)” that allows a robot to safely avoid moving obstacles (security). Further, it is taught an obstacle avoiding robot using “multiple sensors” giving “sensor streams” (at least one sensor data stream) that create “the robot’s observation vector” data (at least partially includes at least one motion vector) including the robot’s current velocity (at least partially include motion vectors extracted from video data) that is used to train a neural network “navigation algorithm”).
Regarding claim 23, the combination of Liang and Shao teach all the claim limitations of claim 13 above; and further teach wherein: the sensor includes at least one vibration sensor that determines measurements of vibrations generated by at least one machine, and wherein the additional sensor data values of the sensor indicate a potential need for maintenance of the at least one machine based on the measured vibrations (Shao, sections 1 and 4.1-4.2 teach using a DBN model to classify communicated motor vibration signals (sensor includes at least one vibration sensor that determines measurements of vibrations generated by at least one machine) for determining the degree and type of a motor issue and effectively diagnose motor failures for better maintenance (additional sensor data values of the sensor indicate a potential need for maintenance of the at least one machine based on the measured vibrations)).
Liang and Shao are combinable for the same rationale as set forth above with respect to claim 13.
Regarding claim 25, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach in response to the at least one sensor data stream falling outside of an expected operating range, receiving, by the device and from the transmitting device, an additional set of machine learning model parameters to the device; and updating, by the device, the predictive machine learning model using the additional set of machine learning model parameters (Liang, sections 3.2-3.4, 4.1, and 5.1 teach “data from the 2-D lidar and the camera are synchronized” since “a sensor’s limited FOV (in response to the at least one sensor data stream falling outside of an expected operating range) can be overcome if another sensor with a high FOV is used in tandem with it (an additional set of machine learning model parameters)”; the camera stream is primarily used for determining “approximate positions, orientations and velocities of all obstacles in the frame” for avoiding the obstacle in a detected sensor FOV before returning to the optimal path; for the processor to train the model via the “training module” on the received sensor vectors on the processor or GPU devices (receiving, by the device and from the transmitting device, by the transmitting device, an additional set of machine learning model parameters to the device; and updating, by the device, the predictive machine learning model using the additional set of machine learning model parameters)).
Regarding claim 26, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach the predictive machine learning model is a probability distribution model, and the probability distribution model is configured to predict average sensor readings of the sensor using the set of machine learning model parameters (Liang, sections 3.2-3.4 and 4.1 teach the POMDP model including “observation space and observation probability distribution given the system state” (predictive machine learning model is a probability distribution model), the observations being “sampled from the system’s state space” sensors, and “output velocity (predict) is sampled from a Gaussian distribution which uses the mean (average sensor readings) and log standard deviation which were updated during training” of the velocity observations (sensor readings of the sensor using the set of machine learning model parameters)).
Regarding claim 28, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach wherein, the set of machine-learned values includes at least one hyperparameter that defines at least one of, an internal organization of the predictive machine learning model, or at least one activation function used by the predictive machine learning model, and the predictive machine learning model is configured by the at least one hyperparameter of the set of machine-learned values to predict the additional sensor data values of the sensor (Liang, sections 3.4-4 teach tuning “hyperparameters” in policy generation for the model and using neural network architecture with ReLU and sigmoid activation functions when predicting the best policy actions based on the sensor inputs).
Regarding claim 29, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach wherein, the set of machine-learned values is determined by the transmitting device during a training of the predictive machine learning model using training data samples of a training data set, and the set of machine-learned values determined by the transmitting device during the training of the predictive machine learning model configure the predictive machine learning model to generate predictions (sections 3.2-3.4, 4.1, and 5.1 teach the training (during training) neural network “navigation algorithm” (predictive machine learning model) with the “the robot’s observation vector” data (training data set) of velocities (set of machine-learned values) by a “training module” (determined by the transmitting device during a training) on the processor or GPU devices, and sections 4.1 and 5.1-5.3 teach creating outputs for the “datasets” in training (predictions)).
However, Liang does not explicitly teach the set of machine-learned values determined by the transmitting device during the training of the predictive machine learning model configure the predictive machine learning model to generate predictions for each training data sample that are respectively consistent with at least one label associated with each training data sample of the training data set.
Shao teaches the set of machine-learned values determined by the transmitting device during the training of the predictive machine learning model configure the predictive machine learning model to generate predictions for each training data sample that are respectively consistent with at least one label associated with each training data sample of the training data set (section 3 teaches DBN model training with labeled training data for tuning weights and biases).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to implement Shao’s teachings of diagnosing motor faults for maintenance into Liang’s teaching of determining robot movement instruction vectors from sensor streams for avoiding moving obstacles and labeled model training in order to “enhance reliability and reduce costs for operation and maintenance of the manufacturing equipment” (Shao, section 1).
Regarding claim 30, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach wherein, the set of machine-learned values is determined by the transmitting device based on the at least one sensor data stream generated by the sensors; and the method further comprises configuring, by the device, the predictive machine learning model using the set of machine-learned values, wherein configuring the predictive machine learning model using the set of machine-learned values enables the device to predict the additional sensor data values of the sensor without requiring the device to determine the set of machine-learned values (Liang, sections 3.2-3.4, 4.1, and 5.1 teach the training (during training) neural network “navigation algorithm” (predictive machine learning model) with the “the robot’s observation vector” data (training data set) of velocities (set of machine-learned values) by a “training module” (determined by the transmitting device during a training) on the processor or GPU devices, and sections 4.1 and 5.1-5.3 teach creating outputs for the “datasets” in training (predictions) and then applying the model to evaluation test sets of sensor data (additional sensor data)).
Regarding claim 31, the combination of Liang and Shao teach all the claim limitations of claim 1 above; and further teach wherein the device predicts additional sensor data values by, configuring, by the device, a set of internal parameters of the predictive machine learning model using the set of machine-learned values, and processing, by the device, the at least one sensor data stream by the predictive machine learning model configured by the set of machine-learned values, wherein the processing causes the predictive machine learning model to predict the additional sensor data values of the sensor (Liang, sections 3.2-3.4, 4.1, and 5.1 teach the training (during training) neural network “navigation algorithm” (predictive machine learning model) with the “the robot’s observation vector” data (training data set) of velocities (set of machine-learned values) by a “training module” (determined by the transmitting device during a training) on the processor or GPU devices, and sections 3.3, 4.1, and 5.1-5.3 teach creating outputs for the “datasets” in training (predictions) and then applying the model to evaluation test sets of sensor data (additional sensor data) that are input and converted to vectors for processing (stream)).
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Liang et al (“Realtime Collision Avoidance for Mobile Robots in Dense Crowds using Implicit Multi-sensor Fusion and Deep Reinforcement Learning”, 2020) hereinafter Liang, in view of Shao et al (“A Deep Learning Approach for Fault Diagnosis of Induction Motors in Manufacturing”, 2017) hereinafter Shao, in view of Radhakrishnan et al (“Inventory Optimization in Supply Chain Management using Genetic Algorithm”, 2009) hereinafter Radhakrishnan.
Regarding claim 9, the combination of Liang and Shao teach all the claim limitations of claim 1 above; however, the combination does not explicitly teach further comprising: detecting, based on the additional sensor data values, a supply shortage of an item; and taking action to avoid running out of the item, wherein the predictive machine learning model predicts stock levels of the item.
Radhakrishnan teaches further comprising: detecting, based on the additional sensor data values, a supply shortage of an item; and taking action to avoid running out of the item, wherein the predictive machine learning model predicts stock levels of the item (sections 3-5 teach using a genetic algorithm (machine learning) to “predict an optimum stock level by using the past records [of detected stock levels] and so that by using the predicted stock level there will be no excess amount of stocks and also there is less means for any shortage”, by “clustering that clusters the stock levels that are either in excess or in shortage and the stock levels that are neither in excess nor in shortage separately”).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify determining robot movement instruction vectors from sensor streams for avoiding moving obstacles, as taught by Liang as modified by diagnosing motor faults for maintenance as taught by Shao, to include using an algorithm for predicting inventory shortages as taught by Radhakrishnan in order to “facilitate the precise determination of the most probable excess stock level and shortage level required for inventory optimization in the supply chain such that minimal total supply chain cost is ensured” (Radhakrishnan, section 5).
Claims 27 and 32-33 are rejected under 35 U.S.C. 103 as being unpatentable over Liang et al (“Realtime Collision Avoidance for Mobile Robots in Dense Crowds using Implicit Multi-sensor Fusion and Deep Reinforcement Learning”, 2020) hereinafter Liang, in view of Shao et al (“A Deep Learning Approach for Fault Diagnosis of Induction Motors in Manufacturing”, 2017) hereinafter Shao, in view of Lajevardi et al (US Pub 20220115148) hereinafter Lajevardi.
Regarding claim 27, the combination of Liang and Shao teach all the claim limitations of claim 1 above; however, the combination does not explicitly teach wherein the predictive machine learning model includes at least one artificial neural network, the set of machine-learned values includes a set of weights and biases, and the set of weights and biases configures an activation of respective neurons of the at least one artificial neural network.
Lajevardi teaches wherein the predictive machine learning model includes at least one artificial neural network (paragraphs 0025 and 0032 teach the “trained machine learning model” being a “neural network”), the set of machine-learned values includes a set of weights and biases, and the set of weights and biases configures an activation of respective neurons of the at least one artificial neural network (paragraph 0025 teach an “device management system 104 sends, at 408, the trained machine learning model to the device 106. Sending the trained machine learning model may involve sending an entire replacement machine learning model, or may instead involve sending updated parameter values for the machine learning model (for example, weights and biases where the model is based on a neural network architecture)”; for processing input sensor data. The weights and biases for tuning internal operations of a model.).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify determining robot movement instruction vectors from sensor streams for avoiding moving obstacles, as taught by Liang as modified by diagnosing motor faults for maintenance as taught by Shao, to include communicating specific updated model parameters for processing sensor data as taught by Lajevardi in order to increase operational accuracy of the trained model outputs (Lajevardi, paragraphs 0025 and 0038-0039).
Regarding claim 32, the combination of Liang and Shao teach all the claim limitations of claim 1 above; however, while the combination teaches a separate dataset of sensor data input to the model after training, the combination does not explicitly teach wherein predicting the additional sensor data values of the sensor using the predictive machine learning model enables the device to use additional sensor data values that are not transmitted to the device by the sensor.
Lajevardi teaches wherein predicting the additional sensor data values of the sensor using the predictive machine learning model enables the device to use additional sensor data values that are not transmitted to the device by the sensor (paragraphs 0025-0026 teach buffering input sensor data in memory (not transmitted to the device by the sensor) before processing with the trained machine learning model on the processor).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify determining robot movement instruction vectors from sensor streams for avoiding moving obstacles, as taught by Liang as modified by diagnosing motor faults for maintenance as taught by Shao, to include communicating specific updated model parameters and routing specifications for processing sensor data as taught by Lajevardi in order to increase operational accuracy of the trained model outputs (Lajevardi, paragraphs 0025 and 0038-0039).
Regarding claim 33, the combination of Liang and Shao teach all the claim limitations of claim 1 above; however, while the combination teaches a utilizing a CPU and GPU for training and evaluation datasets, the combination does not explicitly teach wherein the sensor is a different device than the transmitting device.
Lajevardi teaches wherein the sensor is a different device than the transmitting device (paragraphs 0025-0026 teach buffering input sensor data in memory (not transmitted to the device by the sensor) before processing with the trained machine learning model on the processor).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify determining robot movement instruction vectors from sensor streams for avoiding moving obstacles, as taught by Liang as modified by diagnosing motor faults for maintenance as taught by Shao, to include communicating specific updated model parameters and routing specifications for processing sensor data as taught by Lajevardi in order to increase operational accuracy of the trained model outputs (Lajevardi, paragraphs 0025 and 0038-0039).
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CLINT MULLINAX whose telephone number is 571-272-3241. The examiner can normally be reached on Mon - Fri 8:00-4:30 PT.
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, Alexey Shmatov can be reached on 571-270-3428. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/C.M./Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123