DETAILED ACTION
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on July 23, 2026, has been entered.
Response to Amendment
Claims 1-20 were previously pending. Applicant’s amendment filed July 23, 2026, has been entered in full. Claims 1 and 12 are amended. No claims are added or cancelled. Accordingly, claims 1-20 remain pending.
Response to Arguments
Applicant argues that the amendments to the claims have overcome the previous rejection under 35 U.S.C. 112(b) (Remarks filed July 23, 2026, hereinafter Remarks: Page 7). Examiner agrees. The previous rejection under 35 U.S.C. 112(b) is withdrawn.
Applicant traverses the previous rejections under 35 U.S.C. 103, arguing that the previously-cited ‘Kim’ and ‘Tan’ references do not teach all elements of the amended claims (Remarks: Pages 7-9). For example, Applicant argues that Kim does not teach to “receive a retrained specialized model from the edge server based on the specialized model, wherein the retrained specialized model is associated with an updated confidence greater than the predetermined value” (Remarks: Page 9). Examiner respectfully disagrees. For example, Kim describes an update phase where a retrained specialized model is received from an edge server (e.g., Sec. 3.1; Fig. 2, Update phase), the retrained specialized model having greater than a threshold precision/accuracy (e.g., Fig. 2, Sec. 3.1, especially last sentence). Nevertheless, Examiner has applied a new primary reference – ‘Mondello’ (US 2019/0205765 A1) – in the rejections below because it appears to more-closely match the presently-claimed embodiment.
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, 8-13, and 19-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over ‘Mondello’ (US 2019/0205765 A1) in view of ‘Casey’ (“How Edge Servers Work,” 2021).
Regarding claim 1, Mondello teaches a system for obtaining video analytic output (e.g., Figures 1-2 and 6) comprising:
an edge server (e.g., Fig. 1, central server 101; see Note Regarding Edge Server below); and
a vehicle (e.g., Figs. 1, 2 and 6, vehicle 111 or 113) comprising a controller (Computer 131) configured to:
obtain video data (e.g., [0065], Fig. 3, sensor input 151 includes video data);
reduce a size of the video data to generate reduced video data (e.g., [0066], Fig. 3, step 157, a subset of the video data is selected; Also see, e.g., [0039]-[0042]; The selection of a subset of the video data captured by the sensor is a reduction of size of the video data; I.e., by only selecting sensor data for unknown or poorly classified objects, and not selecting sensor data for known or well-classified objects, the size of the sensor data being selected is reduced from the initial size of the obtained video data; Also note that the claim does not require a particular order of steps – see, e.g., [0078] of the published specification);
determine a confidence of a specialized model by inputting the video data or the reduced video data to the specialized model (e.g., [0061], confidence below a threshold triggers selection of video data; Determining whether confidence is below a threshold includes determination of the confidence; The specialized model is the ANN model used at the vehicle, such as ANN model 119 at Fig. 3; It is specialized at least because it is customized for a particular vehicle ([0082]) and/or specialized through local self-learning ([0085]));
determine that the confidence is less than or equal to a predetermined value (e.g., [0061] – see above);
send at least one of the specialized model or the reduced video data to the edge server in response to the determination that the confidence is less than or equal to the predetermined value (e.g., [0061], the selected video clip of the object with low confidence – i.e., the reduced video data as explained above – is sent to the server when the confidence is below the threshold);
receive a retrained specialized model from the edge server based on the specialized model (e.g., [0063], vehicle receives updated ANN model – i.e., a retrained specialized model based on the specialized model – from the server), wherein the retrained specialized model is associated with an updated confidence greater than the predetermined value (e.g., [0070], updated ANN model [i.e., the retrained specialized model] is capable of recognizing objects that were previously unrecognizable with the original ANN model [i.e., the specialized model]; I.e., the retrained specialized model is associated with an updated confidence greater than the predetermined value because the object is now recognized instead of being unknown or uncertain); and
operate the vehicle using an output of the retrained specialized model (e.g., [0038], Fig. 1, vehicle is operated through steering, braking, and acceleration control according to an output of ANN model 119, which is updated to a retrained specialized model as explained above).
Note Regarding Edge Server. Mondello describes its server as being “centralized” (throughout). A central/centralized server is at the center of a network and not at an edge of the network, and thus is not within the scope of an “edge server” as recited in the claim.
However, Casey does teach edge servers that, rather than being centralized, are located at the edge of a network close to the machines using its applications and data (Page 3). Casey teaches that centralized models are less effective for IoT applications (Page 3), such as smart cars (Page 4), and that edge servers are more effective because their proximity to IoT devices typically provides faster, more reliable user experiences (Page 4). Casey further teaches that edge servers can help deal with the data coming from smart cars (Page 6) and optimize back-and-forth data transmission for AI/ML training (Pages 6-7).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to modify the system of Mondello with the edge server of Casey in order to improve the system with the reasonable expectation that this would result in a system that enjoyed advantages such as faster and more reliable user experiences and/or optimized back-and-forth data transmission for AI/ML training. This technique for improving the system of Mondello was within the ordinary ability of one of ordinary skill in the art based on the teachings of Casey.
Therefore, it would have been obvious to one of ordinary skill in the art to combine the teachings of Mondello and Casey to obtain the invention as specified in claim 1.
Regarding claim 2, Mondello in view of Casey teaches the system according to claim 1, and Mondello further teaches that the specialized model is included in the vehicle (e.g., Fig. 2, ANN model 119 is in vehicle 111; Also see ANN model 119 in vehicle 113 at Fig. 6).
Regarding claim 8, Mondello in view of Casey teaches the system according to claim 1, and Mondello further teaches that the edge server is configured to:
receive the specialized model from the vehicle (e.g., [0120], “In some instances, the result of the unsupervised training (162) is also provided to the centralized server (e.g., 101) to reduce the computation workload of the supervised training (161)”; Note from Fig. 7 that the result of unsupervised training 162 is an updated ANN model 164 that has been specialized through the unsupervised training);
retrain the specialized model received from the vehicle (e.g., [0120], Fig. 7, supervised training 161); and
transmit the retrained specialized model to the vehicle (e.g., [0120], Fig. 7, updated ANN model – i.e., the retrained specialized model – is downloaded to the vehicle from the server).
Regarding claim 9, Mondello in view of Casey teaches the system according to claim 8, and Mondello further teaches that the edge server retrains the specialized model with a general model in the edge server (e.g., [0085], the supervised retraining at the server uses a general global model, as opposed to the specialized models self-learned in each individual vehicle).
Regarding claim 10, Mondello in view of Casey teaches the system according to claim 1, and Mondello further teaches that the edge server is configured to:
determine whether the edge server is available for processing the video data received from the vehicle (e.g., [0049], “Periodically, the server (101) runs the supervised training module (117) to update the ANN model (119).”; During periods where training is performed, the edge server is available for supervised training that includes processing the video data received from the vehicle; During periods where training is not performed, the server is unavailable for training);
store the video data received from the vehicle in a frame buffer in response to determining that the edge server is not available (e.g., [0062], server stores received video data “for the subsequent further training or updating of the ANN model (119) using the supervised training module (117)”; This storage is within the scope of being “within a frame buffer” at least because the video, which is made up of frames, is being saved for later use and thus buffered; also see, e.g., the discussion of memory at [0127] et seq.); and
input the video data to a general model in the edge server to obtain the video analytic output in response to determining that the edge server is available (e.g., [0049], during periods when edge server is available to run supervised training module, the edge server runs supervised training using the video data as a training input to obtain a retrained/updated ANN model).
Regarding claim 11, Mondello in view of Casey teaches the system according to claim 1, and Mondello further teaches that the controller is further configured to:
send the specialized model to the edge server in response to determining that the confidence is less than or equal to the predetermined value (e.g., [0120], output of unsupervised training 162 – i.e., the specialized model – is sent to server for supervised training; e.g., Figs. 3, 4 and 7, training data are transmitted to edge server in response to determining that the confidence is less than the threshold predetermined value; I.e., the purpose of sending the specialized model is to perform supervised training, which itself is performed in response to objects being classified with below-threshold confidence).
Regarding claim 12, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 1. Mondello in view of Casey teaches the system of claim 1 (see above). Accordingly, claim 12 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey for substantially the same reasons as claim 1.
Regarding claim 13, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 2. Mondello in view of Casey teaches the system of claim 2 (see above). Accordingly, claim 13 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey for substantially the same reasons as claim 2.
Regarding claim 19, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 8. Mondello in view of Casey teaches the system of claim 8 (see above). Accordingly, claim 19 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey for substantially the same reasons as claim 8.
Regarding claim 20, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 9. Mondello in view of Casey teaches the system of claim 9 (see above). Accordingly, claim 20 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey for substantially the same reasons as claim 9.
Claim(s) 3, 6-7, 14, and 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey as applied above, and further in view of ‘Kim’ (“Design and Implementation of the Vehicular Camera System using Deep Neural Network Compression,” 2017).
Regarding claim 3, Mondello in view of Casey teaches the system according to claim 1.
Mondello does not teach that a general model in the edge server is compressed to the specialized model. Casey also does not teach this feature.
However, Kim does teach a system that, like Mondello, deploys a trained artificial neural network (ANN) from a server to a vehicle (e.g., Fig. 2). Kim teaches that it is challenging to deploy conventional deep learning ANNs to embedded devices such as vehicles because the network size and amount of computation are too large (e.g., Abstract and Section 1, 2nd-to-last paragraph). Kim solves this problem by compressing a general model (Faster-RCNN) in a server to a specialized model that can be deployed to the vehicle (e.g., Secs. 3.2 and 3.3 describe compression through pruning and quantization; e.g., Fig. 2, compression phase; e.g., Secs. 4.2.5 and 5 explain that the compression provides a smaller and faster model that can be run on a vehicle).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to modify the system of Mondello in view of Casey as applied above with the model compression of Kim in order to improve the system with the reasonable expectation that this would result in a system that trained ANNs that were smaller and faster to execute, and thus were easier to use when embedded in a vehicle. This technique for improving the system of Mondello in view of Casey was within the ordinary ability of one of ordinary skill in the art based on the teachings of Kim.
Therefore, it would have been obvious to one of ordinary skill in the art to combine the teachings of Mondello, Casey and Kim to obtain the invention as specified in claim 3.
Regarding claim 6, Mondello in view of Casey and Kim teaches the system according to claim 3, and Kim further teaches that parameters of the general model are compressed to obtain parameters of the specialized model (e.g., Sec. 3.3, parameters are compressed through conversion from 32-bit real numbers to 8-bit integers).
Regarding claim 7, Mondello in view of Kasey and Kim teaches the system according to claim 3, and Kim further teaches that the general model comprises a machine learning model for processing frames in the video data (e.g., Sec. 2.3, Faster-RCNN).
Regarding claim 14, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 3. Mondello in view of Casey and Kim teaches the system of claim 3 (see above). Accordingly, claim 14 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey and Kim for substantially the same reasons as claim 3.
Regarding claim 17, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 6. Mondello in view of Casey and Kim teaches the system of claim 6 (see above). Accordingly, claim 17 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey and Kim for substantially the same reasons as claim 6.
Regarding claim 18, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 7. Mondello in view of Casey and Kim teaches the system of claim 7 (see above). Accordingly, claim 18 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey and Kim for substantially the same reasons as claim 7.
Claim(s) 4 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey and Kim as applied above, and further in view of ‘Ren’ (“Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” 2016).
Regarding claim 4, Mondello in view of Casey and Kim teaches the system according to claim 3.
Kim’s compressed specialized model recognizes two classes: vehicle sides and rears (Sec. 4.1). Kim teaches using Faster R-CNN as a general model (e.g., Sec. 2.3), but does not explicitly teach a number of classes recognized by Faster-RCNN.
However, Ren does teach details of Faster R-CNN, including that it can recognize 20 (Sec. 4.1, 1st par.) or 80 (Sec. 4.2, 1st par.) different object classes. Accordingly, the disclosure of Ren demonstrates that a number of objects or classes recognized by the compressed specialized model of Kim (2) is less than a number of objects or classes recognized by the Faster R-CNN general model (20 or 80), as required by the claimed invention.
Regarding claim 15, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 4. Mondello in view of Casey, Kim and Ren teaches the system of claim 4 (see above). Accordingly, claim 15 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey, Kim and Ren for substantially the same reasons as claim 4.
Claim(s) 5 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey and Kim as applied above, and further in view of ‘Choi’ (US 2018/0268292 A1).
Regarding claim 5, Mondello in view of Casey and Kim teaches the system according to claim 3.
Kim uses a Faster R-CNN object detector as a general model (e.g., Sec. 2.3) and applies model compression including pruning and quantization to obtain a smaller and faster specialized model (Secs. 3.2-3.3). Kim does not explicitly teach that its compression results in a number of hidden layers of the specialized model being less than a number of hidden layers of the general model. As neither explicitly teach model compression, Mondello and Casey also do not teach this feature
However, Choi does teach an additional technique for compressing a Faster R-CNN object detection model based on knowledge distillation (e.g., Figs. 1 and 4; [0036]). A larger and slower teacher/general model is used to train a smaller and faster compressed/specialized/student model (e.g., [0028]-[0029]). The compressed/specialized/student model has fewer hidden layers than the general/teacher model ([0028], “the teacher model 110 may include a larger number of hidden layers … compared to the student model 120”).
Choi teaches that its techniques “solve the problem of achieving object detection at an accuracy comparable to complex deep learning models, while maintaining speeds similar to a simpler deep learning model” ([0070]). Choi also teaches that “Distillation tends to solve the problem of generalization, in other words, the over-fitting problem” ([0060]).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to modify the system of Mondello in view of Casey and Kim with the knowledge distillation of Choi in order to improve the system with the reasonable expectation that this would result in a system that could solve the problem of achieving object detection at an accuracy comparable to complex deep learning models, while maintaining speeds similar to a simpler deep learning model, and/or that could solve the problem of generalization. This technique for improving the system of Mondello in view of Casey and Kim was within the ordinary ability of one of ordinary skill in the art based on the teachings of Choi.
Therefore, it would have been obvious to one of ordinary skill in the art to combine the teachings of Mondello, Casey, Kim and Choi to obtain the invention as specified in claim 5.
Regarding claim 16, Examiner notes that the claim recites a method that is substantially the same as the method performed by the system of claim 5. Mondello in view of Casey, Kim and Choi teaches the system of claim 5 (see above). Accordingly, claim 16 is also rejected under 35 U.S.C. 103 as being unpatentable over Mondello in view of Casey, Kim and Choi for substantially the same reasons as claim 5.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GEOFFREY E SUMMERS whose telephone number is (571)272-9915. The examiner can normally be reached Monday-Friday, 7:00 AM to 3:30 PM ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Chan Park can be reached at (571) 272-7409. 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.
/GEOFFREY E SUMMERS/Examiner, Art Unit 2669
Prosecution Insights