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 .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 3/23/2026 has been entered.
Response to Arguments
Applicant's arguments filed 3/23/2026 have been fully considered but they are not persuasive.
In regards to the primary thrust of the applicant’s argument in the paragraph between pages 10 and 11 of the remarks, Applicant asserts that Das does not have a “machine learning model distinct from the camera-radar fusion model” and that Das does not disclose “using such a proposed training label to train a second version of a camera-radar fusion model”. Applicant relies in part on paragraph 15 for support for the machine learning model to be distinct from the camera-radar fusion model. Das does disclose what is put forward in paragraph 15. Paragraph 15 discloses that, “In some aspects, an offline machine learning model can be used to process detections made by an online (e.g., deployed on AV) camera-radar fusion model (CRFM).” So, if Das discloses that the machine learning apparatus can be outside of the car, Das would disclose the specifics of that setup. Paragraph 53 along with Figure 2, say that the offline computing devices labelled 214 can perform the processes of the tracking component by themselves. This is further corroborated with paragraphs 39-41 and 46-47. Further, Paragraph 55 of Das discloses that, “As described herein, the localization component 226, the perception component 228, the planning component 230, the tracking component 232, and/or other components of the system 200 may comprise one or more ML models” which more explicitly allows for both the disclosed fusion model to have further separate models and to be further trained. In regards to the second argument presented, Das further discloses the designation of labels, Paragraph 11 of Das discloses that the object classification comprises of labelling the objects by their semantic labels with “an object classification associated with the object (e.g., a vehicle, an oversized vehicle, a pedestrian, a cyclist)”. These constitute semantic labels of the various objects along with claim 106 showing that a ground truth semantic label is used as a comparison during the training process of their disclosed model which results in a further trained version. Applicant’s specification describes that the second version of the camera-radar fusion model or CRFM 324 “may correspond to a new and/or a revised version of CRFM 302 that is undergoing training” in paragraph 53. Applicant’s disclosure allows for the “second version of the model” to merely be the same model that has undergone or is undergoing further training which allows for the training disclosed by Das to read upon the claimed language.
In regards to the first line of argument, where applicant alleges that Das does not disclose a “camera-radar fusion model” or “operations that use camera and radar in a single model to produce detections as claimed”. Paragraph 33 of Das discloses a “hybrid vision-lidar-radar pipeline, and/or the like” with regards to the tracking architecture. As such, applicant’s arguments that no such model is disclosed by Das is not persuasive.
In regards to the second line of argument, applicant alleges that Das does not disclose label generation and that Das does not disclose a separate and distinct machine-learning model from the camera-radar fusion model. On the point of whether Das provides for a distinct machine learning model from the camera-radar fusion model, Das does provide for such a limitation. Firstly, the claim language of, “a machine learning model that is distinct from a camera-radar fusion model” does not require the specific online and offline requirements cited by paragraph 15 of the application’s specification. Even given the broader BRI of the phrasing in the claim, Das does disclose what is put forward in paragraph 15. Paragraph 15 discloses that, “In some aspects, an offline machine learning model can be used to process detections made by an online (e.g., deployed on AV) camera-radar fusion model (CRFM).” So, if Das discloses that the machine learning apparatus can be outside of the car, Das would disclose the specifics of that setup. Paragraph 53 along with Figure 2, say that the offline computing devices labelled 214 can perform the processes of the tracking component by themselves. This is further corroborated with paragraphs 39-41 and 46-47. Further, Paragraph 55 of Das discloses that, “As described herein, the localization component 226, the perception component 228, the planning component 230, the tracking component 232, and/or other components of the system 200 may comprise one or more ML models” which more explicitly allows for both the disclosed fusion model to have further separate models and to be further trained. As such, Das more than discloses the required distinctness from the claims. Das further discloses the designation of labels, Paragraph 11 of Das discloses that the object classification comprises of labelling the objects by their semantic labels with “an object classification associated with the object (e.g., a vehicle, an oversized vehicle, a pedestrian, a cyclist)”. These constitute semantic labels of the various objects along with claim 106 showing that a ground truth semantic label is used as a comparison during the training process. As such, the second argument is not persuasive.
In regards to the third argument, Applicant alleges that Das fails to disclose using the proposed training labels to train a second version of the camera-radar fusion model. Applicant’s specification describes that the second version of the camera-radar fusion model or CRFM 324 “may correspond to a new and/or a revised version of CRFM 302 that is undergoing training” in paragraph 53. Applicant’s disclosure allows for the “second version of the model” to merely be the same model that has undergone or is undergoing further training. The abstract of Das discloses further training the machine learning model. As such, this argument is not persuasive.
As all of the arguments have been considered and found to be not persuasive, the 102 and 103 rejections remain.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1-4, 7-11, 14-17, and 20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Das et al. (US 20210237761 A1).
Claims 1-4. 7-11. 14-17, and 20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Das et al. (US 20210237761 A1), hereinafter referred to as Das.
In regards to claim 1, Das discloses an apparatus comprising: at least one memory comprising instructions (Paragraph 46, Discloses the usage of memories and processors in paragraph 46); and at least one processor comprising a machine learning model that is distinct from a camera-radar fusion model (Paragraphs 21, 33, 39-41, 46-47, 53, and paragraph 55; abstract; and Figure 2, Discloses the usage of memories and processors in paragraph 46 and 21 discloses the usage of machine learning along with the abstract disclosing that a ML or machine learning model is used. Further, Figure 2 visually shows a system where the sensors on the vehicle and the machine learning model are distinct and separate from each other with the model being on the computing device 214 with this being further corroborated by paragraphs 39-41, 47, and 53 and paragraph 33 disclosing a “vision-lidar-radar pipeline” which reads on the camera-radar fusion model with further paragraph 55 disclosing that multiple models can be associated with specific components tied to the pipelines that are distinctly on or off the vehicle), and the at least one processor configured to execute the instructions and cause the at least one processor to: perform, by a first version of the camera-radar fusion model, camera-radar fusion operations that use real time camera sensor data and real time radar sensor data to generate at least one detection (Paragraphs 25 and 23, Paragraph 25 discloses that LIDAR, RADAR, and cameras can be used in conjunction with one another. Further, paragraph 23 discloses the usage of instant techniques which is enough to read upon the real-time requirement), the real time camera sensor data gathered from an environment, and the real time radar sensor data gathered from the environment, the environment comprising objects in a state of change (Paragraphs 25, 11, 23, Paragraph 25 discloses that LIDAR, RADAR, and cameras can be used in conjunction with one another. Paragraph 11 shows an environment where objects are in a state of change by factoring in various aspects of what objects are doing. Further, paragraph 23 discloses the usage of instant techniques which is enough to read upon the real-time requirement); receive, by the machine learning model, the at least one detection generated by the first version of the camera-radar fusion model, wherein the at least one detection corresponds to a scene of the environment at a given time (Paragraphs 25 and 23 and Abstract, Paragraph 25 discloses that LIDAR, RADAR, and cameras can be used in conjunction with one another. Further, paragraph 23 discloses the usage of instant techniques which is enough to read upon the real-time requirement and the abstract discloses that objects are detected via the detection system. Further the detection is received in real time as it is disclosed to be current); receive, by the machine learning model, real time LIDAR sensor data gathered from the environment and corresponding to the scene of the environment at the given time (Paragraphs 25 and 23 and Abstract, Paragraph 25 discloses that LIDAR, RADAR, and cameras can be used in conjunction with one another. Further, paragraph 23 discloses the usage of instant techniques which is enough to read upon the real-time requirement and the abstract discloses that objects are detected via the detection system. Further the detection is received in real time as it is disclosed to be current); wherein real time sensor data comprises the real time camera sensor data and the real time radar sensor data (Paragraphs 25 and 23 and Abstract, Paragraph 25 discloses that LIDAR, RADAR, and cameras can be used in conjunction with one another. Further, paragraph 23 discloses the usage of instant techniques which is enough to read upon the real-time requirement and the abstract discloses that objects are detected via the detection system. Further the detection is received in real time as it is disclosed to be current); wherein the real time sensor data and the real time LIDAR sensor data are configured to enable an autonomous vehicle (AV) to self- navigate through the environment (Paragraphs 27-28 and 4, Paragraphs 27 and 28 go into detail on how the driving apparatus is controlled and how the data collected is used to instruct the vehicle’s movement. Paragraph 4 discloses that the vehicle is autonomous); execute the machine learning model to perform label generation operations on the at least one detection and the real time LIDAR sensor data to generate at least one proposed training label for the at least one detection (Paragraph 11 and paragraph 106, The disclosed classification system classifies various objects with a semantic label being a method of classification disclosed by paragraph 106); and use the at least one proposed training label to train a second version of the camera-radar fusion model (Paragraph 106, Discloses that the first camera-radar fusion model can be further trained based off of the labelling into a second camera-radar fusion model).
In regards to claim 2, Das discloses wherein the at least one processor is further configured to: receive, by the machine learning model, object tracking data corresponding to the scene, wherein the at least one proposed training label for training the second version of the camera-radar fusion model is further based on the object tracking data (Paragraph 36, Discloses that the object detections may associate or use tracking component data for objection classification).
In regards to claim 3, Das discloses wherein the object tracking data is generated by a perception stack of the AV (Paragraph 27, Paragraph 27 describes a perception component that is analogous to the disclosed perception stack).
In regards to claim 4, Das discloses wherein the at least one processor is further configured to: determine a confidence metric that is associated with the at least one proposed training label (Paragraph 73, Discloses that there is a confidence evaluated for the object classification output which would be a label).
In regards to claim 7, Das discloses wherein the first version of the camera-radar fusion operations is performed by the first version of the camera-radar fusion model is-deployed on the AV (Paragraphs 40-41, Paragraph 40 discloses that the vehicle may have all the components of the system on the vehicle with paragraph 41 giving some examples), and wherein the real time camera sensor data is captured by a camera sensor on the AV (Paragraphs 40-41, Paragraph 40 discloses that all sensors may be on the vehicle with paragraph 41 explicitly disclosing via example that the camera is on the vehicle), the real time radar sensor data is captured by a radar sensor on the AV (Paragraphs 40-41, Paragraph 40 discloses that all sensors may be on the vehicle with paragraph 41 explicitly disclosing via example that the radar is on the vehicle), and the LIDAR sensor data is captured by a LIDAR sensor on the AV (Paragraphs 40-41, Paragraph 40 discloses that all sensors may be on the vehicle with paragraph 41 explicitly disclosing that one of the sensors included on the vehicle can be lidar sensors).
Claims 8 and 15 are similar to claim 1, and they are similarly rejected.
Claims 9 and 16 are similar to claim 2, and they are similarly rejected.
Claim 10 is similar to claim 3, and it is similarly rejected.
Claims 11 and 17 are similar to claim 4, and they are similarly rejected.
Claims 14 and 20 are similar to claim 7, and they are similarly rejected.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
The factual inquiries 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.
Claims 5, 12, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Das et al. (US 20210237761 A1), hereinafter referred to as Das, in view of Marvasti et al. (US 20250166352 A1), hereinafter referred to as Marvasti.
In regards to claim 5, Das does not explicitly disclose the elements of this claim.
Marvasti is within the same field as Das as Marvasti is similarly based around the usage of multiple machine learning systems that deal with a variety of data points to provide input for a vehicle.
However, Marvasti discloses wherein the at least one processor is further configured to: discard the at least one proposed training label when the confidence metric is less than a first threshold confidence level (Paragraph 27, Discloses discarding a hypothesis which would read on a label if the confidence was low enough).
It would have been prima facie obvious to combine the teachings of Marvasti with the teachings of Das. It would have led to a predictable increase in accuracy for the labels overall. Being able to discard labels that have low confidence values would allow for the system to be able to regulate what labels can be applied, and it would allow it to effectively keep bad or poor labels away from any future training which would ensure that the next training would be more accurate. As such, it would have been prima facie obvious to combine the teachings of the two arts.
Claims 12 and 18 are similar to claim 5, and they are similarly rejected.
Claims 6, 13, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Das et al. (US 20210237761 A1), hereinafter referred to as Das, in view of Song et al. (US 20190220525 A1), hereinafter referred to as Song.
In regards to claim 6, Das does not disclose the elements of this claim.
Song is within the same field of the art as Das as both are directed towards machine learning and training set ups utilizing multiple datasets.
However, Song does disclose wherein the at least one processor is further configured to: designate the at least one proposed training label for review by a labeler when the confidence metric is less than a second threshold confidence level (Paragraph 94, Discloses manual review by a human labeler if the confidence level is below a certain threshold).
It would have been prima facie obvious to combine the teachings of Das and Song. The combination of the two would have led to a predictable increase in the accuracy of the labels. Having an external reviewer to handle certain kinds of low confidence figures would increase accuracy as an external reviewer can catch when a label was accurate but with a low confidence rating or when a label is low confidence due to being inapplicable. As such, using one here would have increased the accuracy of the labels for future training, and it would have been prima facie obvious to do so.
Claims 13 and 19 are similar to claim 6, and they are similarly rejected.
Response to Amendment
The amendment, entered 3/23/2026, has been considered in full. The amended language overcomes the 112 rejections. However, the amended language after further consideration does not overcome the respective 102 rejections and 103 rejections. Further, all of the claims have been either amended previously or with this action, however, Claim 13 is described as “(Original)” in this amendment even though the term, “training” has been added to the claim language. Examiner asks that any future response to this action designate Claim 13 as “(Previously Presented)” or “(Currently Amended)” to be in line with the other amended claims and in line with any future amendments or lack thereof.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Bhargava et al. (US 20220284598 A1) is a pertinent piece of prior art that focuses on 3D object tracking using semantic key points that can be utilized with autonomous vehicles.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CONOR AIDAN O'MALLEY whose telephone number is (571)272-0226. The examiner can normally be reached Monday - Friday 9:00 am. - 5:00 pm. EST.
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, Andrew Moyer can be reached at 5722729523. 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.
CONOR AIDAN. O'MALLEY
Examiner
Art Unit 2675
/CONOR A O'MALLEY/ Examiner, Art Unit 2675
/GREGORY A MORSE/ Supervisory Patent Examiner, Art Unit 2698
Prosecution Insights