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 .
Response to Amendment
This action is in response to amendments filed on 07/01/2026, in which claims 1, 5, 7-9, 13, and 15-17 are amended. Claims 21-23 are added. Claims 1-23 are pending.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 07/21/2026 is being considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1:
Step 1: Claim 1 recites a method which falls into the statutory category of process.
Step 2A Prong 1: Claim 1 recites multiple mental processes: reducing a dimensionality of the data acquired in step a) by passing the data to the encoder to produce a compressed latent representation of the data, wherein the encoder extracts task-specific features from raw sensor data and discards non-relevant data to reduce dimensionality of the data; and determining a driving operation by the policy head directly outputting a driving control decision using the compressed latent representation of the data-[.]];. These limitations are directed to abstract ideas merely using a generic tool, policy head or encoder, to perform a concept which can be performable in the human mind.
Step 2A prong 2: Claim 1 does not integrate the abstract idea into a practical application since the additional elements of:
Acquiring data related to a task for operation a vehicle wherein the task is selected out of lane centering and lane changing, is insignificant extra-solution activity: data gathering.
training a deep learning model using the data acquired, wherein the deep learning model includes an encoder and a policy head for the task, is merely an instruction to apply the abstract idea using a generic computer component.
controlling the vehicle based on the driving control decision, wherein the controlling comprises at least one of vehicle direction control, power train control or brake control, is merely an instruction to apply the abstract idea.
wherein when the task is lane centering the operation comprises lane centering and when the task is lane changing the operation comprises lane changing, merely links the abstract idea to a field of use.
Step 2B: Claim 1 does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional element of acquiring data Is considered insignificant extra-solution activity because it is well-understood, routine, conventional activity as evidenced by MPEP §2106.05(d)(II)(I). Meanwhile, training a deep learning model, including a policy head and encoder, using the data acquired is the use of a generic computer component to apply the abstract idea given it merely cites a generic deep learning model with generic components that encode and produce outputs see MPEP 2106.05(f). Furthermore, controlling the vehicle based on the decision determined by the model is also considered mere instructions to apply the abstract idea given it merely explains the general way you apply the abstract idea: to control a vehicle. Finally, limitations that specify the task being lane centering or changing merely link the abstract idea to a field of use see MPEP 2106.05(h). Claim 1 is not patent eligible.
Regarding claim 2, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, wherein the data acquired includes recorded human driving data from a same or a separate vehicle.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 2 is not patent eligible.
Regarding claim 3, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, wherein the data acquired includes artificially augmented data.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 3 is not patent eligible.
Regarding claim 4, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, wherein the data is acquired using a sensor of a same or a separate vehicle, the sensor including one or more lidar sensors, radar sensors, infrared sensors, and/or image sensors.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 4 is not patent eligible.
Regarding claim 5, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, wherein the step c) further includes applying a mask that is element-wise multiplied by the compressed latent representation to further reduce the
Regarding claim 6, the rejection of claim 5 is incorporated, further the claim recites “The method of claim 5, further comprising normalizing mask values.” This limitation amounts to more specifics of the judicial exception. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 6 is not patent eligible.
Regarding claim 7, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, further comprising applying a loss function to evaluate a difference between the driving operation determined by the policy head and a driving operation benchmark and updating parameters of the encoder and the compressed latent representation based on a gradient of the loss function to achieve improved driving operation decisions.” This limitation amounts to mathematical calculations. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 7 is not patent eligible.
Regarding claim 8, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, further comprising identifying, by a data sharing module, overlapping per-entity elements between a first sparse representation associated with the task and a second sparse representation associated with another task; creating a shared latent representation based on the overlapping per-entity elements; and configuring one or more overlapping elements of the compressed latent representation produced by a first encoder of a first deep learning model, such that the compressed latent representation is configured to be shareable by a second encoder of a second deep learning model.” This limitation recites a mental process: identifying overlapping elements between representations and creating a shared representation based on the overlapping elements. Furthermore, the additional elements of the data sharing module and configuring one or more overlapping elements of the compressed latent representation produced by a first encoder of a first deep learning model, such that the compressed latent representation is configured to be shareable by a second encoder of a second deep learning model amount to mere instructions to apply the abstract idea by a generic computer component and insignificant extra-solution activity given it’s merely the transmitting of data from one encoder to another and therefore is considered well-known, routine, conventional as evidenced by MPEP §2106.05(d)(II)(I). As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 8 is not patent eligible.
Regarding claim 9:
Step 1: Claim 9 recites a method which falls into the statutory category of process.
Step 2A Prong 1: Claim 9 recites multiple mental processes: obtaining a compressed latent representation of the data acquired in step a), wherein the compressed latent representation is produced by an encoder that extracts task-specific features from raw sensor data and discards non-relevant data to reduce dimensionality of the data; and determining a driving operation by the policy head directly outputting a driving control decision using the compressed latent representation of the data-[.]];. These limitations are directed to abstract ideas merely using a generic tool, policy head or encoder, to perform a concept which can be performable in the human mind.
Step 2A prong 2: Claim 9 does not integrate the abstract idea into a practical application since the additional elements of:
Acquiring data related to a task for operation a vehicle, is insignificant extra-solution activity: data gathering.
operating a deep learning model with the data acquired, wherein the deep learning model includes a policy head for the task, is merely an instruction to apply the abstract idea using a generic computer component.
controlling the vehicle based on the driving control decision, wherein the controlling comprises at least one of vehicle direction control, power train control or brake control, is merely an instruction to apply the abstract idea.
wherein when the task is lane centering the operation comprises lane centering and when the task is lane changing the operation comprises lane changing, merely links the abstract idea to a field of use.
Step 2B: Claim 9 does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional element of acquiring data Is considered insignificant extra-solution activity because it is well-understood, routine, conventional activity as evidenced by MPEP §2106.05(d)(II)(I). Meanwhile, operating a deep learning model, including a policy head, using the data acquired is the use of a generic computer component to apply the abstract idea given it merely cites a generic deep learning model with generic components that encode and produce outputs see MPEP 2106.05(f). Furthermore, controlling the vehicle based on the decision determined by the model is also considered mere instructions to apply the abstract idea given it merely explains the general way you apply the abstract idea: to control a vehicle. Finally, limitations that specify the task being lane centering or changing merely link the abstract idea to a field of use see MPEP 2106.05(h). Claim 9 is not patent eligible.
Regarding claim 10, the rejection of claim 9 is incorporated, further the claim recites “The method of claim 9, wherein the data acquired includes recorded human driving data from a same or a separate vehicle.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 10 is not patent eligible.
Regarding claim 11, the rejection of claim 9 is incorporated, further the claim recites “The method of claim 9, wherein the data acquired includes artificially augmented data.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 11 is not patent eligible.
Regarding claim 12, the rejection of claim 9 is incorporated, further the claim recites “The method of claim 9, wherein the data is acquired using a sensor of a same or a separate vehicle, the sensor including one or more lidar sensors, radar sensors, infrared sensors, and/or image sensors.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 12 is not patent eligible.
Regarding claim 13, the rejection of claim 9 is incorporated, further the claim recites “The method of claim 9, wherein the step c) further includes applying a mask that is element-wise multiplied by the compressed latent representation to further reduce the integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 13 is not patent eligible.
Regarding claim 14, the rejection of claim 13 is incorporated, further the claim recites “The method of claim 13, further comprising normalizing mask values.” This limitation amounts to more specifics of the judicial exception. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 14 is not patent eligible.
Regarding claim 15, the rejection of claim 9 is incorporated, further the claim recites “The method of claim 9, further comprising applying a loss function to evaluate a difference between the driving operation determined by the policy head and a driving operation benchmark and updating parameters of the encoder and the compressed latent representation based on a gradient of the loss function to achieve improved driving operation decisions.” This limitation amounts to mathematical calculations. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 15 is not patent eligible.
Regarding claim 16, the rejection of claim 9 is incorporated, further the claim recites “The method of claim 9, further comprising identifying, by a data sharing module, overlapping per-entity elements between a first sparse representation associated with the task and a second sparse representation associated with another task; creating a shared latent representation based on the overlapping per-entity elements; and configuring one or more overlapping elements of the compressed latent representation produced by a first encoder of a first deep learning model, such that the compressed latent representation is configured to be shareable by a second encoder of a second deep learning model.” This limitation recites a mental process: identifying overlapping elements between representations and creating a shared representation based on the overlapping elements. Furthermore, the additional elements of the data sharing module and configuring one or more overlapping elements of the compressed latent representation produced by a first encoder of a first deep learning model, such that the compressed latent representation is configured to be shareable by a second encoder of a second deep learning model amount to mere instructions to apply the abstract idea by a generic computer component and insignificant extra-solution activity given it’s merely the transmitting of data from one encoder to another and therefore is considered well-known, routine, conventional as evidenced by MPEP §2106.05(d)(II)(I). As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 16 is not patent eligible.
Regarding claim 17:
Step 1: Claim 17 recites a method which falls into the statutory category of process.
Step 2A Prong 1: Claim 17 recites multiple mental processes: identifying one or more overlapping elements between the data related to the task and a compressed latent representation related to another task, wherein the compressed latent representation is produced by a second deep learning model with a second encoder…. and determining a driving operation by the policy head using the compressed latent representation of the data. These limitations are directed to abstract ideas merely using a generic tool, policy head or encoder, to perform the concept which can be performable in the human mind.
Step 2A prong 2: Claim 17 does not integrate the abstract idea into a practical application since the additional elements of:
Acquiring data related to a task for operation a vehicle, is insignificant extra-solution activity.
training a deep learning model using the data acquired, wherein the deep learning model includes an encoder and a policy head for the task, is merely an instruction to apply the abstract idea using a generic computer component
the compressed latent representation is configured to be shareable with the first encoder of the first deep learning model, is insignificant extra-solution activity.
controlling the vehicle based on the driving control decision, wherein the controlling comprises at least one of vehicle direction control, power train control or brake control, is merely an instruction to apply the abstract idea.
wherein when the task is lane centering the operation comprises lane centering and when the task is lane changing the operation comprises lane changing, merely links the abstract idea to a field of use.
Step 2B: Claim 17 does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional element of acquiring data Is considered insignificant extra-solution activity because it is well-understood, routine, conventional activity as evidenced by MPEP §2106.05(d)(II)(I). Meanwhile, training a deep learning model using the data acquired is so that a generic computer component can apply the abstract idea and mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Additionally, the limitation of “the compressed latent representation is configured to be shareable with the first encoder of the first deep learning model” amounts to insignificant extra-solution activity given it’s merely the transmitting of data from one encoder to another and therefore is considered well-known, routine, conventional as evidenced by MPEP §2106.05(d)(II)(I). Furthermore, controlling the vehicle based on the decision determined by the model is also considered mere instructions to apply the abstract idea given it merely explains the general way you apply the abstract idea: to control a vehicle. Finally, limitations that specify the task being lane centering or changing merely link the abstract idea to a field of use see MPEP 2106.05(h). Claim 17 is not patent eligible.
Regarding claim 18, the rejection of claim 17 is incorporated, further the claim recites “The method of claim 17, wherein the data acquired includes recorded human driving data from a same or a separate vehicle.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 18 is not patent eligible.
Regarding claim 19, the rejection of claim 17 is incorporated, further the claim recites “The method of claim 17, wherein the data acquired includes artificially augmented data.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 19 is not patent eligible.
Regarding claim 20, the rejection of claim 17 is incorporated, further the claim recites “The method of claim 17, wherein the data is acquired using a sensor of a same or a separate vehicle, the sensor including one or more lidar sensors, radar sensors, infrared sensors, and/or image sensors.” This limitation amounts to more specifics of the additional element of acquiring data which was already deemed insignificant extra-solution activity. As such the claim does not have any additional elements that amount to an integration of the judicial exception into a practical exception, nor to significantly more than the judicial exception. Claim 20 is not patent eligible.
Regarding claim 21, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, wherein the compressed latent representation comprises a one-dimensional vector representing a plurality of elements of the data, each element associated with a different feature of surroundings of the vehicle, the different features including one or more of lane borderlines, lane centerlines, vehicles nearby, traffic signs, and tree contours”. This limitation amounts to more specifics of the abstract idea of producing a compressed latent representation. Claim 21 is not patent eligible.
Regarding claim 22, the rejection of claim 21 is incorporated, further the claim recites “The method of claim 21, wherein each element of the plurality of elements is a single or multi-dimensional matrix”. This limitation amounts to more specifics of the abstract idea of producing a compressed latent representation. Claim 22 is not patent eligible.
Regarding claim 23, the rejection of claim 1 is incorporated, further the claim recites “The method of claim 1, wherein the task comprises traffic sign reading”. This limitation amounts to merely linking the abstract idea to a field of use see MPEP 2106.05(h). Claim 23 is not patent eligible.
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 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.
Claim(s) 1-7, 9-15, 21-23 are rejected under 35 U.S.C. 103 as being unpatentable over Bojarski (US20200324795A1) in view of Cultrera (Explaining Autonomous Driving by Learning End-to-End Visual Attention).
Regarding claim 1, Bojarski teaches a) acquiring data related to a task for operating a vehicle (Abs, The ground truth data may be associated with the training sensor data for training one or more deep neural networks (DNNs) to compute outputs corresponding to autonomous machine operations); wherein the task is selected out of lane centering and lane changing (Par 0045, the compute trajectories for a vehicle in deployment may also more closely follow centers or rails of lanes of travel. In addition, as a trajectory may include changing lanes, making turns, and/or the like, the adjustments to the ego-trajectories for ground truth generation may also enable the turns, lane changes, and/or other maneuvers of the computed trajectories of the DNN 116 to shift from centers or rails of lanes of travel to centers or rails of other lanes of travel); b) training a deep learning model using the data acquired, wherein the deep learning model includes an encoder and a policy head for the task (Abs, The ground truth data may be associated with the training sensor data for training one or more deep neural networks (DNNs) to compute outputs corresponding to autonomous machine operations, Par 0034, For example, and without limitation, the DNN 116 may include any type of machine learning model……neural networks (e.g., auto-encoders, convolutional, recurrent, perceptrons, long/short term memory/LSTM, Hopfield, Boltzmann, deep belief, deconvolutional, generative adversarial, liquid state machine, etc.)); d) determining a driving operation by the policy head directly outputting a driving control decision using the compressed latent representation of the data-[.]]; and e) controlling the vehicle based on the driving control decision, wherein when the task is lane centering the operation comprises lane centering and when the task is lane changing the operation comprises lane changing; wherein the controlling comprises at least one of vehicle direction control, power train control or brake control (Par 0045, Where these ego-trajectories are used directly to train the DNN 116 to compute future trajectories for a vehicle, the compute trajectories in deployment may similarly include shifts from rail or center of the lane of travel. As such, by augmenting the ego-trajectory using the map data to generate ground truth data that more closely follow a center or rail of the lane of travel, the compute trajectories for a vehicle in deployment may also more closely follow centers or rails of lanes of travel. In addition, as a trajectory may include changing lanes, making turns, and/or the like, the adjustments to the ego-trajectories for ground truth generation may also enable the turns, lane changes, and/or other maneuvers of the computed trajectories of the DNN 116 to shift from centers or rails of lanes of travel to centers or rails of other lanes of travel.)
Bojarski fails to teach c) reducing a dimensionality of the data acquired in step a) by passing the data to the encoder to produce a compressed latent representation of the data, wherein the encoder extracts task-specific features from raw sensor data and discards non-relevant data to reduce dimensionality of the data.
Cultrera teaches c) reducing a dimensionality of the data acquired in step a) by passing the data to the encoder to produce a compressed latent representation of the data, wherein the encoder extracts task-specific features from raw sensor data and discards non-relevant data to reduce dimensionality of the data (Table 1 Convolution backbone architecture, Section 3.3, The first part of the model takes care of extracting features from input images….This is made possible using a fully convolutional structure, table 1 shows dimensionality reduction).
Bojarski and Cultrera are analogous to the claimed invention because they are both in the field of autonomous vehicles. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the end-to-end attention system of Cultrera within the DNN of Bojarski to extract features from images and understand which pixels contribute more to a decision while preserving spatial information (Cultrera Section 3.3).
Regarding claim 2, Bojarski teaches the data acquired includes recorded human driving data from a same or a separate vehicle (Par. 0045, As such, the ground truth generator 112 may generate a final ground truth trajectory—or data representative thereof, such as points along a polyline defining the final ground truth trajectory…. For example, and with reference to FIG. 3, because the ego-trajectory 202 may be generated based on a driven path of a human driver, the ego-trajectory may stray from a center or rail of the lane of travel…. Where these ego-trajectories are used directly to train the DNN…..).
Regarding claim 3, Bojarski teaches wherein the data acquired includes artificially augmented data (Par. 0023, in addition, in some examples, the sensor data 102 may be used within the process 100 without any pre-processing (e.g., in a raw or captured format), while in other examples,
the sensor data 102 may undergo pre-processing (e.g., noise balancing, demosaicing, scaling, cropping, augmentation…).
Regarding claim 4, Bojarski teaches wherein the data is acquired using a sensor of a same or a separate vehicle, the sensor including one or more lidar sensors, radar sensors, infrared sensors, and/or image sensors (Par. 0003, In such instances, conventional approaches leverage on-board sensors of the vehicles—such as vision sensors (e.g., cameras, LIDAR, RADAR, etc.)—to detect objects, road features (e.g., lane markings, road edges, etc.), free-space boundaries, wait condition information, intersection structure and pose, and/or the like. For example, deep neural networks (DNNs) may be leveraged to process the sensor data from the on-board sensors and compute outputs corresponding to any of the above operations.)
Regarding claim 5, Cultrera teaches, which Bojarski fails to teach, wherein the step c) further includes applying a mask that is element-wise multiplied by the compressed latent representation to further reduce the complexity dimensionality of the data acquired in step a) (Figure 4, Attention Block. A weight vector α is generated by a linear layer with SoftMax activation. The final descriptor is obtained as a weighted sum of RoI-pooled features, attention block is the mask).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have used the visual attention system in Cultrera along with the system of Bojarski to “focus on salient regions of the image” because “in an autonomous driving context, explicitly modeling saliency can help the network to take decisions considering relevant objects or environmental cues, such as other vehicles or intersections” (Cultrera Section 3.5).
Regarding claim 6, Cultrera teaches, which Bojarski fails to teach, further comprising normalizing mask values (Figure 4, Attention Block. A weight vector α is generated by a linear layer with SoftMax activation. The final descriptor is obtained as a weighted sum of RoI-pooled features, SoftMax is normalizing).
Regarding claim 7, Cultrera teaches, which Bojarski fails to teach, further comprising applying a loss function to evaluate a difference between the driving operation determined by the policy head and a driving operation benchmark, and updating parameters of the encoder and the compressed latent representation based on a gradient of the loss function to achieve improved driving operation decisions (Section 3.7, For each image, a steering angle s represents the action performed by the driver. Therefore the loss function is calculated using Mean Squared Error (MSE) between the predicted value and the ground truth steer (4)…. Where sGT and sp represent respectively ground truth and predicted steer values. The model is trained using the Adam optimizer [25] for 20 epochs with batch size of 64 and initial learning rate of 0,0001, final portion implies gradient descent and updating parameters of the CNN given the system is trained end-to-end).
Regarding claims 9-15, the claims cite essentially the same limitations as claims 1-7 but slightly broader and as such are rejected for the same reasons stated above.
Regarding claim 21, Cultrera teaches, which Bojarski fails to teach, wherein the compressed latent representation comprises a one-dimensional vector representing a plurality of elements of the data, each element associated with a different feature of surroundings of the vehicle, the different features including one or more of lane borderlines, lane centerlines, vehicles nearby, traffic signs, and tree contours (Fig 2, Section 3.5, At first, region features ri obtained by RoI-pooling are flattened and concatenated together in a single vector r, ROI pooling creates a version of the compressed latent representation comprising the single vector r, Fig 3, features include lane markings and tree contours which are used in the ROI pooling).
Regarding claim 22, Cultrera teaches, which Bojarski fails to teach, wherein each element of the plurality of elements is a single or multi-dimensional matrix (Fig 2, the RoI pooling layer creates a vector whose elements are multi-dimensional as shown in the figure).
Regarding claim 23, Bojarski teaches wherein the task comprises traffic sign reading (Par 0029, As a further example, where the DNN 116 is trained to generate outputs corresponding to intersection (e.g., bounding shape vertices corresponding to a bounding shape encompassing an intersection), the correlator 110 may determine each of the features of the intersection (e.g., traffic lights, traffic signs, labels or markings on the driving surface, etc.) that correspond to an intersection such that the ground truth generator 112 may generate bounding shapes that encompass each of the features (e.g., similar to visualization 240 of FIG. 2B).)
Claim(s) 8, 16, 17-20 are rejected under 35 U.S.C. 103 as being unpatentable over Bojarski (US20200324795A1) in view of Cultrera (Explaining Autonomous Driving by Learning End-to-End Visual Attention) as applied to claims 1 and 19 above, and further in view of Bilen (Integrated Perception with Recurrent Multi-Task Neural Networks).
Regarding claim 8, the combination of Bojarski and Cultrera teaches the method of claim 1 but fails to teach further comprising identifying, by a data sharing module, overlapping per-entity elements between a first sparse representation associated with the task and a second sparse representation associated with another task; creating a shared latent representation based on the overlapping per-entity elements; and configuring one or more overlapping elements of the compressed latent representation produced by a first encoder of a first deep learning model, such that the compressed latent representation is configured to be shareable by a second encoder of a second deep learning model.
Bilen teaches further comprising identifying, by a data sharing module, overlapping per-entity elements between a first sparse representation associated with the task and a second sparse representation associated with another task; creating a shared latent representation based on the overlapping per-entity elements; and configuring one or more overlapping elements of the compressed latent representation produced by a first encoder of a first deep learning model, such that the compressed latent representation is configured to be shareable by a second encoder of a second deep learning model (Fig 1, We propose a modular multi-task architecture in which several perceptual tasks are integrated in a synergistic manner. The subnetwork φ0 enc encodes the data x0 (an image in the example) producing a representation h shared between K different tasks. Each task estimates one of K different labels xα (object class, location, and parts in the example) using K decoder functions ψα dec. Each task contributes back to the shared representation by means of a corresponding encoder function φα enc. The loop is closed in a recurrent configuration by means of suitable integrator functions (not shown here to avoid cluttering the diagram).
Bilen is analogous to the claimed invention because it is in the field of multi-task neural networks. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to have modified the combination of Bojarski and Cultrera to use the Multinet system in Bilen to engage in multiple tasks while sharing data between multiple encoders to improve training on image data (Bilen Abs, In this manner, we show that the performance of individual tasks in standard benchmarks can be improved first by sharing features between them and then, more significantly, by integrating their solutions in the common representation).
Regarding claim 16, the claim cites essentially the same limitation as claim 8 but slightly broader and as such is rejected for the same reasons stated above.
Regarding claim 17, the claim is essentially a combination of claims 1 and 8 and as such the rejections of claims 1 and 8 are incorporated and the claim is rejected for the same reasons stated above.
Regarding claim 18, Bojarski teaches wherein the data acquired includes recorded human driving data from a same or a separate vehicle (Par. 0045, As such, the ground truth generator 112 may generate a final ground truth trajectory—or data representative thereof, such as points along a polyline defining the final ground truth trajectory…. For example, and with reference to FIG. 3, because the ego-trajectory 202 may be generated based on a driven path of a human driver, the ego-trajectory may stray from a center or rail of the lane of travel…. Where these ego-trajectories are used directly to train the DNN…..).
Regarding claim 19, Bojarski teaches wherein the data acquired includes artificially augmented data (Par. 0023, in addition, in some examples, the sensor data 102 may be used within the process 100 without any pre-processing (e.g., in a raw or captured format), while in other examples,
the sensor data 102 may undergo pre-processing (e.g., noise balancing, demosaicing, scaling, cropping, augmentation…).
Regarding claim 20, Bojarski teaches wherein the data is acquired using a sensor of a same or a separate vehicle, the sensor including one or more lidar sensors, radar sensors, infrared sensors, and/or image sensors (Par. 0003, In such instances, conventional approaches leverage on-board sensors of the vehicles—such as vision sensors (e.g., cameras, LIDAR, RADAR, etc.)—to detect objects, road features (e.g., lane markings, road edges, etc.), free-space boundaries, wait condition information, intersection structure and pose, and/or the like. For example, deep neural networks (DNNs) may be leveraged to process the sensor data from the on-board sensors and compute outputs corresponding to any of the above operations.)
Response to Arguments
Applicant is reminded that all business is required to be done in writing (see MPEP 713.04). In the response filed 7/1/2026, Applicant makes reference to agreements alleged to have been made regarding art rejections by the examiner during an interview. The applicant may have intended to rely on the interview agenda as their official response to the previous Office action, which is not proper.
Furthermore, the applicant does not provide a clear argument regarding the art rejections in the applicant’s arguments filed with the response to the previous Office action.
Nevertheless, Applicants’ arguments with respect to the prior art rejections (102 and 103) of the independent claims, including those made in the agenda, have been considered but are moot because they do not apply to the new reference Cultrera.
Applicant's arguments regarding the 35 U.S.C. 101 rejections of claims 1-20 filed 07/01/2026 have been fully considered but they are not persuasive. The applicant alleges the added details in the amended independent claims provide better linkage between specific tasks and the control of the vehicle and as such overcome the 35 U.S.C 101 rejection. The examiner respectfully disagrees. The new details added about the encoder are not sufficient to make the encoder more than a generic computer component which merely applies the judicial exception see MPEP 2106.05(f) given it merely describes the basic features of a generic encoder, that it reduces the dimensionality of data by extracting features from data. Additionally, the new added details about outputting a driving control decision in limitation d) of claim 1 are not sufficient to make the determining step more than a mental process given that a human being can still do the step with the aid of pen and paper to output the decision. Finally, limitation e) does not integrate the judicial exception into a practical application because controlling the vehicle using the determined decision is considered mere instructions to apply the abstract idea given it merely explains the general way you apply the abstract idea: to control a vehicle. The remaining portions of limitation e) merely describe the field of use of the judicial exception: lane centering and changing.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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.
/NATNAEL A ASEGDEW/Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
