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 Amendment
The amendment filed 4/28/2026 has been entered. Claims 1, 3-12 and 14-20 remain pending in the application and claims 2 and 13 have been cancelled. Applicant’s amendments to the claims have overcome each and every rejection under 35 U.S.C. 112(b) and 35 U.S.C. 101 set forth in the Non-Final Office Action mailed 1/28/2026. Applicant’s amendments to the claims and drawings have additionally overcome each and every objection set forth in the Non-Final Office Action mailed 1/28/2026.
Response to Arguments
Applicant argues that the Examiner’s assertion that creating partitioned datasets through the invention of Shen is improper, and contradictory to Shen’s stated purpose. Further, Applicant asserts that the claimed subject matter explicitly recites “generating complete, segregated datasets (e.g., a first for autonomous control data, a second for manual control data) and passing those discrete datasets to processors to independently train specific ML models”. Using this claim, Applicant finally argues that modifying Shen to “compile and store massive, segregated datasets for individual ML training models would directly frustrate Shen’s core objective”.
The distinction as defined by Applicant for what the claimed invention is intended to perform is much appreciated, and is helpful for advancing a common understanding. However, in light of the claim language as recited, the arguments are not persuasive. The underlined portions above in Applicant’s argument directly contradict broadest reasonable interpretation of the claimed limitations. Despite Applicant’s assertions, there is no “explicit recitation” that the datasets are “complete” or that they are “massive”. Currently, the amended independent claims recite four steps: collecting one or more sensor datapoints, filtering each sensor datapoint based on two categories--- autonomous control data or manual control data, generating at least one dataset based on the filtered data, and training at least one machine learning model based on at least one of the datasets.
What Shen lacks, as indicated in the Non Final Office Action, is simply the explicit recitation of generating at least one of a first dataset or a second dataset, based on the filtered data. Shen is clearly performing the partitioning of the data, as evidenced in Col 3, Line 14: “differentiating between machine decision (as calculated by the advanced AI chipset based on the raw video and raw sensor data), and human judgement (captured as driver behavior data by sensors).” One of ordinary skill in the art would thus understand in the realm of data processing, the system of Shen already partitions the data into separate datasets, but simply lacks explicit recitation of doing so.
Finally, addressing the amended limitation, the first machine learning model is intended to “optimize a performance of the one or more vehicle functions” and the second machine learning model is intended to be used to “update a map database”. Shen explicitly aims to achieve similar goals in Col 3, Line 64: “The computer system (via one of the processing units) and method provides the detected data as input for a training model, the provided input being used by the training model to produce decision logic for a self-driving vehicle. In some embodiments, the providing of the map input includes transmitting the provided model input to cloud service, wherein training applications access the provided model input from the cloud service”.
Ultimately, Applicant’s arguments are not persuasive in light of the recited claimed limitations, as a broadest reasonable interpretation of the claims does not support the proposed distinctions over the prior art.
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claim(s) 1, 3-4, 7-12, 14, and 17-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shen et al., hereinafter Shen (Document ID: US 10803324 B1).
Regarding claims 1, 12, and 20, Shen reaches a computer-implemented method, a system, and a non-transitory computer readable storage medium including one or more sequences of one or more instructions which, when executed by one or more processors comprise/comprising:
collecting one or more sensor datapoints corresponding to one or more vehicle functions, wherein each sensor datapoint of the one or more sensor datapoints comprises an autonomous functionality status identifier of the respective vehicle function of the one or more vehicle functions (see at least Col 8, Line 29: “a differentiator module 290, which identifies critical data that indicates a differentiation between the captured driving data from the set of sensors 120 and captured human driver behavior data from the set of sensors 130”. See also FIG. 2 which shows the process of parsing data between machine decisions and human decisions for vehicle functions, wherein the labels act as an autonomous functionality status identifier);
filtering each sensor data point into one of autonomous control data or manual control data to generate filtered data, based on the corresponding autonomous functionality status identifier of the respective sensor data point (see at least Col 7, Line 31: “one of the processing units 140, 150 may be configured with a software module (logic/algorithm) to derive “human decisions” from the human driver's behavior data. The processing units 140, 150 may be further configured to derive “machine decisions” from the captured video images and driving data”);
Shen does not explicitly teach
generating at least one of a first dataset or a second dataset, based on the filtered data, wherein the first dataset corresponds to the autonomous control data, and wherein the second dataset corresponds to the manual control data;
But Shen does teach in at least Col 7, Line 45: “the software modules may then compare the derived “human decisions” and “machine decisions” and only transport differential data (i.e., data showing differences between “human decisions” and “machine decisions”) to the cloud service 180.” Shen additionally specifies in Col 8, Line 46 “Note, the system platform 100 (AI processing unit/module 150 and primary data unit/module 160) may also transmit the machine decisions 215 and human decisions 225 without being first differentiated (in cases where this includes a manageable quantity of images/data”.
Therefore it would have been obvious to one of ordinary skill in the art before the filing date of the claimed invention to have modified the cloud service data storage of Shen with a partitioned dataset using the filtered data, including one for autonomous control data or a second for manual control data in order to execute a design choice for how the preprocessed vehicle data is stored using the cloud service. Shen is principally concerned with reducing the amount of unnecessary data that is transmitted and stored to a cloud service (or to other powerful servers) in Col 2, Line 32. Therefore, it would have been obvious to one of ordinary skill in the art to modify the dataset generated by Shen to differentiate between the autonomous control and manual control data after performing the preprocessing step, as the amount of data transmitted need not be any different by splitting the dataset.
In view of the modification, Shen further teaches
controlling at least one of: a first processor for training, using the first dataset, a first machine learning (ML) model to optimize a performance of the one or more vehicle functions; or a second processor for training, using the second dataset, a second machine learning (ML) model to update a map database (see at least Col 2, Line 56: “learning and testing system platform for self-driving vehicles, which generates the output used to improve the reliability and intelligence of the decision logic for the self-driving vehicles.” Which uses “embedded central processing unit (CPU)”. See also Col 3, line 24: “The cloud service, as part of these embodiments, stores and organizes the real-time map information and makes it readily accessible for the map applications and self-driving applications. The cloud service may also store, organize, and make available geo-based driving information including road information as used in maps, drivers' behavioral information”, which the cloud service necessarily utilizes a processor for).
Regarding claim 12 specifically, Shen additionally teaches
a memory configured to store computer executable instructions (memory 90)
one or more processors (processor 50)
Regarding claims 3 and 14, modified Shen teaches the method of claim 1 and the system of claim 12, and in view of the modification Shen further teaches
determining driver behavior data of a driver of at least one vehicle, based on the second dataset (see at least Col 3, line 42: “driver behavior data representing actions by a human controlling the vehicle”, which uses the vehicle sensors).
Regarding claims 4 and 19, modified Shen teaches the method of claim 1 and the system of claim 12, and in view of the modification Shen further teaches
a first processor to optimize a performance of the one or more vehicle functions, based on the first dataset, wherein the first processor utilizes one or more of first heuristic algorithms, first statistical analysis or first user inputs corresponding to manual review of the first dataset to optimize the performance of the one or more vehicle functions (see at least Col 8, Line 33: “An AI model being learned and trained for determining the self-driving decision logic/algorithm by a training application can be continuously updated from the images/data transferred and stored at the cloud service 180”. Essentially the combination above establishes that Shen collects the data that makeup the first and second databases, and here Shen is using a heuristic algorithm in the form of an AI model that continuously aims to make “more intelligent human-like driving decisions”, which is considered more optimal performance ); or
a second processor to update a map database, based on the second dataset, wherein the second processor utilizes one or more of second heuristic algorithms, second statistical analysis or second user inputs corresponding to manual review of the second dataset to update the map database.
Regarding claims 7 and 17, modified Shen teaches the method of claim 1 and the system of claim 12, and in view of the modification, Shen further teaches
the second dataset comprises a functionality identifier of at least one vehicle function of the one or more vehicle functions executed under manual mode, one or more timestamps corresponding to at least one of a start time of execution of the at least one vehicle function or an end time of execution of the at least one vehicle function, one or more location identifiers corresponding to at least one of the start time of execution of the at least one vehicle function or the end time of execution of the at least one vehicle function (see at least Col 4, Line 14: “the captured driver behavior data” which corresponds to a functionality identifier of at least one vehicle function executed under manual mode, “a timestamp for the parsed information” which would necessarily correspond to at least one of a start time of execution of the at least one vehicle function or an end time of execution of the at least one vehicle function and “GPS data” as a location identifier that can me matched to at least one of the start time of execution of the at least one vehicle function or the end time of execution of the at least one vehicle function).
Regarding claims 8 and 18, modified Shen teaches the method of claim 1 and the system of claim 12, and Shen further teaches
the one or more sensor datapoints corresponding to the one or more vehicle functions are collected from one or more sensors comprising at least one of a camera, a lidar, a millimeter wave radar, a GPS, or an inertia measurement unit (IMU), a wheel speed sensor, a speed sensor, an acceleration sensor, a steering angle sensor, a rain sensor, or an ambient light sensor (see at least Col 1, Line 62: “HD camera, radar, sonar, global positioning system (GPS), and Lidar sensors of the self-driving vehicle.” For some sensor types).
Regarding claim 9, modified Shen teaches the method of claim 1, and Shen further teaches
the one or more vehicle functions comprise at least one of: one or more vehicle maneuver functions, or one or more vehicle auxiliary functions (see at least Col 5, Line 48: “actions the human driver makes in controlling the vehicle at any moment.”).
Regarding claim 10, modified Shen teaches the method of claim 9, and Shen further teaches
the one or more vehicle maneuver functions comprise at least one of: cruise control functions, braking functions, lane assist functions, steering functions, gear shift functions using stick shift and clutch, paddle shifters, or automatic transmission, or driving functions (see at least Col 5, Line 50: “how hard the driver steps on the break (break pressure, break percentage), how hard the driver steps on the acceleration paddle (throttle percentage), gear chosen,”).
Regarding claim 11, modified Shen teaches the method of claim 10, and Shen further teaches
the one or more vehicle auxiliary functions comprise at least one of: headlight control functions, wiper control functions, heating, ventilation and air conditioning (HVAC) control functions, window defrost function, or SOS functions (see at least Col 5, Line 54: “indication signals, and whether the windshield wiper blade is on or off.”).
Claim(s) 5-6 and 15-16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shen in view of Stenneth (Document ID: US 20160247394 A1).
Regarding claims 5 and 15, modified Shen teaches the method of claim 1 and the system of claim 12, and Shen further teaches in at least Col 4, Line 14: “a timestamp for the parsed information” and “GPS data” as a location identifier, wherein the parsed data is synchronized with the sensor datapoints as captured driving data. But Shen does not explicitly teach a functionality identifier of a respective vehicle function of the one or more vehicle functions.
each sensor datapoint of the one or more sensor datapoints further comprises a functionality identifier of a respective vehicle function of the one or more vehicle functions, one or more timestamps associated with the respective vehicle function of the one or more vehicle functions, and one or more location identifiers associated with the respective vehicle function of the one or more vehicle functions (see at least Col 4, Line 14: “a timestamp for the parsed information”, “GPS data” as a location identifier).
Instead, Stenneth, whose invention pertains to classifying vehicle based on autonomy level teaches in P [0053] the ability to “extract relevant features or patterns for each type of vehicle (i.e. manual, partially autonomous, and fully autonomous) from the sensor data.”
It would have been obvious to one of ordinary skill in the art before the filing date of the claimed invention to have modified the timestamp and GPS data associated with driving data of Shen with the autonomy level feature detection of Stenneth in order to successfully identify patterns for each type of vehicle control mode from sensor data. As identified by Stenneth in P [0001], the classification is important for "traffic safety analysis, resource allocation, road infrastructure management, and other applications". It is important to note that in the background section Shen provides context for the challenges to self-driving vehicles, including limited test coverage, high data volume, and evolving models. Shen provides a forward looking system for allowing real time comparison between a manual driving decision process and an autonomous decision process. Therefore, it would have been obvious to one of ordinary skill in the art before the filing date of the invention to modify the system of Shen by applying it to vehicles that do perform autonomous driving as in the system of Stenneth.
Regarding claims 6 and 16, modified Shen teaches the method of claim 1 and the system of claim 12, and in view of the modification, Shen further teaches in at least Col 4, Line 14: “a timestamp for the parsed information” which would necessarily correspond to at least one of a start time of execution of the at least one vehicle function or an end time of execution of the at least one vehicle function and “GPS data” as a location identifier that can me matched to at least one of the start time of execution of the at least one vehicle function or the end time of execution of the at least one vehicle function. But Shen does not explicitly teach that the first dataset comprises a functionality identifier of at least one vehicle function of the one or more vehicle functions executed under autonomous mode,
Instead, Stenneth teaches in P [0042] “the classification platform 109 may process sensor data received from one or more vehicles to determine vehicle status information, driver condition information, environmental information, or a combination thereof.” And in P [0043] Stenneth relates the classification to a database system: “the database 111 may store classification features (e.g., vehicle status feature, a driver condition feature, an environmental feature, derived feature) for one or more vehicles.”
It would have been obvious to one of ordinary skill in the art before the filing date of the claimed invention to have modified the timestamp and GPS data associated with driving data of Shen with the autonomy level feature detection and database of Stenneth in order to successfully identify patterns for an autonomous vehicle control mode from sensor data. As identified by Stenneth in P [0001], the classification is important for "traffic safety analysis, resource allocation, road infrastructure management, and other applications".
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 nonprovisional extension fee (37 CFR 1.17(a)) 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.
Additional art made of record and not relied upon is considered pertinent to applicant's disclosure.
Document ID: US 20220171401 A1
Invention pertains to comparing human and machine based driving for training better autonomous driving models.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Dairon Estevez whose telephone number is (703)756-4552. The examiner can normally be reached M-F 8:00AM - 4:00PM.
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, Khoi Tran can be reached at (571) 272-6919. 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.
/D.E./Examiner, Art Unit 3656
/KHOI H TRAN/Supervisory Patent Examiner, Art Unit 3656
Prosecution Insights