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 .
Priority
Acknowledgement is made of Applicant’s claim of priority from 63556717, filed February 22, 2024. However, the subject matter of claims 1-42 are not adequately supported by provisional application No. 63/556,717; therefore, claims 1-42 are not entitled to the benefit of the provisional filing date and are accorded the filing date of the present non-provisional application.
Information Disclosure Statement
The information disclosure statements (“IDS”) filed on October 14, 2024 were reviewed and the listed references were noted.
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 25-30 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. The claim(s) does/do not fall within at least one of the four categories of patent eligible subject matter because “a machine-readable medium having stored thereon…” could refer to transitory forms of signal transmission (i.e., “signals per se”) which does not fall under any of the statutory categories (see MPEP 2106.03(I)).
Claim interpretation affects the evaluation of both criteria for eligibility. For example, in Mentor Graphics v. EVE-USA, Inc., 851 F.3d 1275, 112 USPQ2d 1120 (Fed. Cir. 2017), claim interpretation was crucial to the court’s determination that claims to a "machine-readable medium" were not to a statutory category. In Mentor Graphics, the court interpreted the claims in light of the specification, which expressly defined the medium as encompassing "any data storage device" including random-access memory and carrier waves. Although random-access memory and magnetic tape are statutory media, carrier waves are not because they are signals similar to the transitory, propagating signals held to be non-statutory in Nuijten. 851 F.3d at 1294, 112 USPQ2d at 1133 (citing In re Nuijten, 500 F.3d 1346, 84 USPQ2d 1495 (Fed. Cir. 2007)). Accordingly, because the BRI of the claims covered both subject matter that falls within a statutory category (the random-access memory), as well as subject matter that does not (the carrier waves), the claims as a whole were not to a statutory category and thus failed the first criterion for eligibility.
The rejection of claims 25-30 may be overcome by amending the claim to, for example, recite as: “a non-transitory machine-readable medium having stored thereon…”.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 5, 13, 17, 25, 31 and 37 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024).
Regarding claim 1, Chen teaches a processor (Chen, Para. [0034], the processor 12) comprising:
one or more arithmetic logic units (ALUs) configured to perform (Chen, Para. [0034], the processor may be a central processing unit (CPU)), using one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks), seed trajectory generation for optimization-based motion planning, the one or more neural networks comprising:
an observation encoder (Chen, Para. [0037], the generator network includes an encoder) configured to:
a noise prediction network (Chen Para. [0079], the decoder is an LSTM model), configured to:
generate a plurality of first trajectories based on the representation of the initial state, the representation of the target state, and random noise (Chen Para. [0079], the decoder is an LSTM model and generates the predicted trajectory of each target object according to the second trajectory information (i.e., representation of initial state), the third trajectory information (i.e., representation of the target state), and the noise).
Although Chen teaches generating trajectories based on an initial state representation, a target state representation, and noise (Chen Para. [0079]), Chen does not explicitly teach “receive a representation of an environment”, “generate, by encoding the representation of the environment, an environment embedding” and " generate a plurality of seed trajectories by denoising, in parallel, the plurality of first trajectories based on the environment embedding”. However, in an analogous field of endeavor, Doersch teaches the system can receive the input image (Doersch, Para. [0028]). The system processes the input image using an image encoder neural network to generated an encoded representation of the input image (i.e., environment embedding) (Doersch, Para. [0079]). Doersch further teaches at each reverse diffusion iteration, the system processes an input for the reverse diffusion iteration that includes the noisy point trajectories as of the reverse diffusion iteration using the trajectory diffusion neural network and conditioned on the encoded representation of the input image to generate a denoising output that defines an update to the noisy point trajectories (Doersch, Para. [0094]).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to modify the processor of Chen with the teachings of Doersch by including encoding an input image representing the environment and generating seed trajectories by denoising the noisy point trajectories based on the encoded representation of the image. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for object tracking across many possible realistic future trajectories, as recognized by Doersch. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Although Chen in view of Doersch teaches generating trajectories based on an initial state representation, a target state representation, and noise (Chen Para. [0079]), they do not explicitly teach “receive the environment embedding, a representation of an initial state, and a representation of a target state”. However, in an analogous field of endeavor, Hughes teaches the labeled event data (i.e., environment embedding) and one or more vectors (i.e., a representation of an initial state and a representation of a target state) may be input into a diffusion model (Hughes, Para. [0176]). The one or more vector may include at least one of an agent two dimensional coordinates on a sporting event's field, an agent position, an agent team, an indicator indicating the agent is a ball, or player visibility information (Hughes, Para. [0175]). The event encoder may embed the event data, embedding the event data further including: tokenzing the labeled event data using a linear projection; applying sinusoidal positional embeddings to specify temporal occurrences of the event data; processing the event data with stacked encoders; and outputting event embeddings (Hughes, Para. [0177]).GuH
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date to modify the processor of Chen in view of Doersch with the teachings of Hughes by including that the diffusion model receives an environment embedding, a representation of an initial state, and a representation of a target state (i.e., the one or more vectors). One having ordinary skill in the art before the effective filing date of the claimed invention would have been motivated to combine these references because doing so would allow for accurately predicting trajectory, as recognized by Hughes. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Regarding claim 5, Chen in view of Doersch further in view of Hughes teaches the processor according to claim 1,
wherein the environment embedding comprises an environment vector (Doersch, Para. [0079]-[0080], the system processes the input image using an image encoder neural network to generate an encoded representation of the input image. The encoded representation includes a respective feature vector),
wherein the noise prediction network comprises a multi-layer perceptron (MLP) (Chen, Para. [0043], a multilayer perceptron (MLP)), the MLP being configured to:
receive the representation of the initial state and the representation of the target state (Hughes, Para. [0176], teaches the labeled event data (i.e., environment embedding) and one or more vectors (i.e., a representation of an initial state and a representation of a target state) may be input into a diffusion model), and
encode the representation of the initial state and the representation of the target state to generate a trajectory endpoints vector (Hughes, Para. [0177], the tracking decoder may use attention to embed and fuse the one or more vectors with the event embeddings), and
wherein the noise prediction network is configured to concatenate the environment vector and the trajectory endpoints vector to generate a concatenated environment vector (Hughes, Para. [0177], the tracking decoder may use attention to embed and fuse the one or more vectors with the event embeddings).
The proposed combination as well as the motivation for combining the Chen, Doersch and Hughes references presented in the rejection of Claim 1, apply to Claim 5 and are incorporated herein by reference. Thus, the processor recited in Claim 5 is met by Chen in view of Doersch further in view of Hughes.
Claims 13 and 17 recite systems with elements corresponding to the processors recited in Claims 1 and 5, respectively. Therefore, the recited elements of these claims are mapped to the proposed combination in the same manner as the corresponding elements in their corresponding processor claims. Additionally, the rationale and motivation to combine the Chen, Doersch and Hughes references, presented in rejection of Claim 1, apply to these claims. Finally, the combination of the Chen, Doersch and Hughes references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claim 25 recites a computer-readable storage medium storing a program with instructions corresponding to the elements recited in Claim 1. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch and Hughes references, presented in rejection of Claim 1, apply to this claim. Finally, the combination of Chen, Doersch and Hughes references discloses a computer readable storage medium (Doersch, Para. [0225], the computer storage medium can be a machine-readable storage device).
Claim 31 recites a robot with elements corresponding to the elements recited in Claim 1. Therefore, the elements of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch and Hughes references, presented in rejection of Claim 1, apply to this claim. Finally, the combination of Chen, Doersch and Hughes references discloses one or more processors (Chen, Para. [0034], the processor 12), one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks) and a robot (Chen, Para. [0040], a mobile robot).
Claim 37 recites a method with steps corresponding to the elements of the processor recited in Claim 1. Therefore, the recited steps of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding system claim. Additionally, the rationale and motivation to combine the Chen, Doersch and Hughes references, presented in rejection of Claim 1, apply to this claim.
Claims 2, 14, 26, 32 and 38 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024), as applied to claims 1, 5, 13, 17, 25, 31 and 37 above, and further in view of Achim et al. (US 12,579,725 B1) and Huapeng Su (US 2023/0196670 A1).
Regarding claim 2, Chen in view of Doersch further in view of Hughes teaches the processor according to claim 1, as described above.
Although Chen in view of Doersch further in view of Hughes teaches an encoder (Chen, Para. [0037]), they do not explicitly teach “wherein the observation encoder comprises a vision transformer”. However, in an analogous field of endeavor, Achim teaches the predictive model for normalized depth prediction can be implemented using a variety of architectures, such as a U-Net architecture, temporal models, vision transformers, Res-Nets (Residual Networks), CNNs (Convolutional Neural Networks), etc. (Achim, Col. 24, lines 32-36).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to modify the processor of Chen in view of Doersch further in view of Hughes with the teachings of Achim by including that the observation encoder is a vision transformer. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for estimating properties of objects, as recognized by Achim.
Although Chen in view of Doersch further in view of Hughes and Achim teaches the image capturing device may be a depth sensor (Chen, Para. [0040]), they do not explicitly teach “wherein the representation of the environment comprises depth image data and camera properties corresponding to the depth image data”. However, in an analogous field of endeavor, Su teaches various types of AR data may be derived from various raw data inputs, including RGB images, camera intrinsics and/or camera transforms, 3D feature points, and/or depth images from a depth sensor (LiDAR, stereo camera, etc.), among other types of possible data. Camera intrinsics can include various known or readily determined properties of camera 104, such as focal length, aperture, optical center, angle of view, focal point, etc. For example, knowing the focal point of a camera can allow a rough approximation of distance (depth) to a feature when that feature is in focus (Su, Para. [0040]).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date to modify the processor of Chen in view of Doersch further in view of Hughes and Achim with the teachings of Su by including that the representation of the environment includes depth image data and camera properties. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for estimating positions in 3D space, as recognized by Su. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Claim 14 recites a system with elements corresponding to the processor recited in Claim 2. Therefore, the recited elements of this claim are mapped to the proposed combination in the same manner as the corresponding element in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch and Hughes references, presented in rejection of Claim 2, apply to this claim. Finally, the combination of the Chen, Doersch and Hughes references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claim 26 recites a computer-readable storage medium storing a program with instructions corresponding to the elements recited in Claim 2. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes, Achima and Su references, presented in rejection of Claim 2, apply to this claim. Finally, the combination of Chen, Doersch, Hughes, Achima and Su references discloses a computer readable storage medium (Doersch, Para. [0225], the computer storage medium can be a machine-readable storage device).
Claim 32 recites a robot with elements corresponding to the elements recited in Claim 2. Therefore, the elements of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes, Achima and Su references, presented in rejection of Claim 2, apply to this claim. Finally, the combination of Chen, Doersch, Hughes, Achima and Su references discloses one or more processors (Chen, Para. [0034], the processor 12), one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks) and a robot (Chen, Para. [0040], a mobile robot).
Claim 38 recites a method with steps corresponding to the elements of the processor recited in Claim 2. Therefore, the recited steps of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding system claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes, Achima and Su references, presented in rejection of Claim 2, apply to this claim.
Claims 3-4 and 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024), Achim et al. (US 12,579,725 B1) and Huapeng Su (US 2023/0196670 A1), as applied to claims 2, 14, 26, 32 and 38 above, and further in view of Hisako Sugano (US 2021/0134049 A1).
Regarding claim 3, Chen in view of Doersch further in view of Hughes, Achim and Su teaches the processor according to claim 2, wherein the observation encoder comprises:
a first network configured to encode the depth image data (Chen, Para. [0037], the generator network includes an encoder).
Although Chen in view of Doersch further in view of Hughes, Achim and Su teaches an encoder (Chen, Para. [0037]), they do not explicitly teach “a second network configured to encode the camera properties” and “wherein the observation encoder is configured to generate the environment embedding based on the encoded depth image data and the encoded camera properties”. However, in an analogous field of endeavor, Sugano teaches the encoding device generates an encoded stream by encoding the camera parameters, the two-dimensional image data, the depth data, and the shadow maps supplied from the conversion device (Sugano, Para. [0072]).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date to modify the processor of Chen in view of Doersch further in view of Hughes, Achim and Su with the teachings of Sugano by including encoding camera properties and generating an encoded stream (i.e., environment embedding) by encoding the camera parameters and depth data. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for encoding camera parameter information with depth data for processing, as recognized by Sugano. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Regarding claim 4, Chen in view of Doersch further in view of Hughes, Achim and Su teaches the processor according to claim 2, wherein the observation encoder comprises a multi-layer perceptron (MLP) (Chen, Para. [0043], a multilayer perceptron (MLP)).
Although Chen in view of Doersch further in view of Hughes, Achim and Su teaches an encoder (Chen, Para. [0037]), they do not explicitly teach to “receive, as input, the camera properties corresponding to the depth image data”, “encode the camera properties to generate a vector representation of the camera properties” and “wherein the observation encoder is configured to generate the environment embedding by encoding the depth image data and the vector representation of the camera properties to generate an environment vector”. However, in an analogous field of endeavor, Sugano teaches the encoding device generates an encoded stream by encoding the camera parameters, the two-dimensional image data, the depth data, and the shadow maps supplied from the conversion device (Sugano, Para. [0072]).
The proposed combination as well as the motivation for combining the Chen, Doersch, Hughes, Achim, Su and Sugano references presented in the rejection of Claim 3, apply to Claim 4 and are incorporated herein by reference. Thus, the processor recited in Claim 4 is met by Chen in view of Doersch further in view of Hughes, Achim, Su and Sugano.
Claims 15-16 recite systems with elements corresponding to the processors recited in Claims 3-4, respectively. Therefore, the recited elements of these claims are mapped to the proposed combination in the same manner as the corresponding elements in their corresponding processor claims. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes, Achima, Su and Sugano references, presented in rejection of Claim 3, apply to these claims. Finally, the combination of the Chen, Doersch, Hughes, Achima, Su and Sugano references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claims 6, 18, 27, 33 and 39 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024), as applied to claims 1, 5, 13, 17, 25, 31 and 37 above, and further in view of Vaezi Joze et al. (US 2019/0294871 A1).
Regarding claim 6, Chen in view of Doersch further in view of Hughes teaches the processor according to claim 1, wherein the noise prediction network is configured to generate the plurality of first trajectories based on the representation of the initial state, the representation of the target state, and the random noise by:
encoding the representation of the initial state and the representation of the target state with time steps to generate a problem embedding (Hughes, Para. [0097], the event data’s data stream includes the event’s timestamp, 2D coordinates, agent-type, and event category).
The proposed combination as well as the motivation for combining the Chen, Doersch and Hughes references presented in the rejection of Claim 1, apply to Claim 6 and are incorporated herein by reference.
Although Chen in view of Doersch further in view of Hughes teaches determining trajectories based on an initial state representation, a target state representation, and noise (Chen Para. [0079]), they do not explicitly teach “generating a plurality of noise tensors of predefined dimensions, each noise tensor of the plurality of noise tensors corresponding to a first trajectory of the plurality of first trajectories” and “generating each first trajectory of the plurality of first trajectories based on the problem embedding and a respective noise tensor of the plurality of noise tensors”. However, in an analogous field of endeavor, Vaezi Joze teaches the skeleton generator network may use a “U” shape architecture that receives the action label (i.e., problem embedding) and optionally, a random noise vector, where the output of the skeleton generator network may be a predetermined tensor representing the skeleton trajectories having a number of time-steps. The random noise vector (e.g., an 8x1x128 tensor) may be provided for each time step (Vaezi Joze, Para. [0036]).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date to modify the processor of Chen in view of Doersch further in view of Hughes with the teachings of Vaezi Joze by including determining the first trajectories based on the problem embedding and the noise tensor for each time step. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for estimating a trajectory, as recognized by Vaezi Joze. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Claim 18 recites a system with elements corresponding to the processor recited in Claim 6. Therefore, the recited elements of this claim are mapped to the proposed combination in the same manner as the corresponding element in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Vaezi Joze references, presented in rejection of Claim 6, apply to this claim. Finally, the combination of the Chen, Doersch, Hughes and Vaezi Joze references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claim 27 recites a computer-readable storage medium storing a program with instructions corresponding to the elements recited in Claim 6. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Vaezi Joze references, presented in rejection of Claim 6, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Vaezi Joze references discloses a computer readable storage medium (Doersch, Para. [0225], the computer storage medium can be a machine-readable storage device).
Claim 33 recites a robot with elements corresponding to the elements recited in Claim 6. Therefore, the elements of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Vaezi Joze references, presented in rejection of Claim 6, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Vaezi Joze references discloses one or more processors (Chen, Para. [0034], the processor 12), one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks) and a robot (Chen, Para. [0040], a mobile robot).
Claim 39 recites a method with steps corresponding to the elements of the processor recited in Claim 6. Therefore, the recited steps of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding system claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Vaezi Joze references, presented in rejection of Claim 6, apply to this claim.
Claims 7, 9, 19, 21, 28, 34 and 40 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024), as applied to claims 1, 5, 13, 17, 25, 31 and 37 above, and further in view of Kingma et al. (US 2023/0267315 A1).
Regarding claim 7, Chen in view of Doersch further in view of Hughes teaches the processor according to claim 1, wherein the noise prediction network comprises a diffusion model (Hughes, Para. [0071], the transformer neural network may further include a diffusion model), and
wherein the denoising, in parallel, the plurality of first trajectories based on the environment embedding comprises:
providing, to (Doersch, Para. [0094], at each reverse diffusion iteration, the system processes an input for the reverse diffusion iteration that includes the noisy point trajectories as of the reverse diffusion iteration using the trajectory diffusion neural network and conditioned on the encoded representation of the input image to generate a denoising output that defines an update to the noisy point trajectories)
denoising, by (Doersch, Para. [0094], at each reverse diffusion iteration, the system processes an input for the reverse diffusion iteration that includes the noisy point trajectories as of the reverse diffusion iteration using the trajectory diffusion neural network and conditioned on the encoded representation of the input image to generate a denoising output that defines an update to the noisy point trajectories).
The proposed combination as well as the motivation for combining the Chen, Doersch and Hughes references presented in the rejection of Claim 1, apply to Claim 7 and are incorporated herein by reference.
Although Chen in view of Doersch further in view of Hughes teaches determining a denoised trajectory using a diffusion model (Doersch, Para. [0094]), they do not explicitly teach “instantiating a plurality of instances of the diffusion model, each instance of the diffusion model corresponding to a first trajectory of the plurality of first trajectories”. However, in an analogous field of endeavor, Kingma teaches the user computing device can implement multiple parallel instances of a single diffusion model (Kingma, Para. [0044]).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to modify the processor of Chen in view of Doersch further in view of Hughes with the teachings of Kingma by including instantiating a plurality of instances of the diffusion model for determining a trajectory. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for removing noise using a diffusion model, as recognized by Kingma. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filling date.
Regarding claim 9, Chen in view of Doersch further in view of Hughes and Kingma teaches the processor according to claim 7, wherein the diffusion model is a conditional diffusion model having a UNet architecture (Doersch, Para. [0053], the trajectory diffusion neural network can be a two-dimensional convolutional neural network, e.g., one that has a U-Net or other convolutional architecture).
The proposed combination as well as the motivation for combining the Chen, Doersch, Hughes and Kingma references presented in the rejection of Claim 7, apply to Claim 9 and are incorporated herein by reference. Thus, the processor recited in Claim 9 is met by Chen in view of Doersch further in view of Hughes and Kingma.
Claims 19 and 21 recite systems with elements corresponding to the processors recited in Claims 7 and 9, respectively. Therefore, the recited elements of these claims are mapped to the proposed combination in the same manner as the corresponding elements in their corresponding processor claims. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Kingma references, presented in rejection of Claim 7, apply to these claims. Finally, the combination of the Chen, Doersch, Hughes and Kingma references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claim 28 recites a computer-readable storage medium storing a program with instructions corresponding to the elements recited in Claim 7. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Kingma references, presented in rejection of Claim 7, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Kingma references discloses a computer readable storage medium (Doersch, Para. [0225], the computer storage medium can be a machine-readable storage device).
Claim 34 recites a robot with elements corresponding to the elements recited in Claim 7. Therefore, the elements of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Kingma references, presented in rejection of Claim 7, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Kingma references discloses one or more processors (Chen, Para. [0034], the processor 12), one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks) and a robot (Chen, Para. [0040], a mobile robot).
Claim 40 recites a method with steps corresponding to the elements of the processor recited in Claim 7. Therefore, the recited steps of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding system claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Kingma references, presented in rejection of Claim 7, apply to this claim.
Claims 8 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024) and Kingma et al. (US 2023/0267315 A1), as applied to claims 7, 9, 19, 21, 28, 34 and 40 above, and further in view of Qi et al. (US 12,602,851 B1).
Regarding claim 8, Chen in view of Doersch further in view of Hughes and Kingma teaches the processor according to claim 7, as described above.
Although Chen in view of Doersch further in view of Hughes and Kingma teaches implementing parallel instances of a diffusion model (Kingma, Para. [0044]), they do not explicitly teach “wherein the denoising, by each instance of the diffusion model based on the environment embedding, the corresponding first trajectory is performed iteratively, wherein an input to a subsequent iteration of the denoising operation is generated based on an output of a previous iteration of the denoising operation”. However, in an analogous field of endeavor, Qi teaches a diffusion model may perform an iterative denoising process in which, at each timestep of a set of successive timesteps, the amount of noise in the preceding image is reduced in a manner based on the third text input. After a selected number of iterations or a selected condition has been met, sufficient noise may be removed from the initial image to generate an output image (Qi, Col. 7 line 38 - Col. 8 line 14).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date to modify the processor of Chen in view of Doersch further in view3 of Hughes and Kingma with the teachings of Qi by including iteratively denoising using a diffusion model, wherein the input to a subsequent iteration of the denoising operation is generated based on an output of the previous iteration. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for iteratively removing noise using a diffusion model, as recognized by Qi. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Claim 20 recites a system with elements corresponding to the processor recited in Claim 8. Therefore, the recited elements of this claim are mapped to the proposed combination in the same manner as the corresponding element in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes, Kingma and Qi references, presented in rejection of Claim 8, apply to this claim. Finally, the combination of the Chen, Doersch, Hughes, Kingma and Qi and references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claims 10, 22, 29, 35 and 41 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024), as applied to claims 1, 5, 13, 17, 25, 31 and 37 above, and further in view of Zeng et al. (2021/0149404 A1).
Regarding claim 10, Chen in view of Doersch further in view of Hughes teaches the processor according to claim 1, as described above.
Although Chen in view of Doersch further in view of Hughes teaches determining denoised (i.e., seed) trajectories (Doersch, Para. [0094]), they do not explicitly teach “wherein the one or more ALUs are further configured to perform, by an optimization-based motion planner, motion planning, wherein the optimization-based motion planner is configured to: receive, as input, the plurality of seed trajectories” and “generate, by performing optimization on the plurality of seed trajectories, a final motion plan from the initial state to the target state”. However, in an analogous field of endeavor, Zeng teaches the motion planning module can generate a plurality of candidate vehicle trajectories for the autonomous vehicle, generate, using the one or more machine-learned models, a cost value for each candidate trajectory in the plurality of candidate vehicle trajectories for the autonomous vehicle, and determine a motion plan for the autonomous vehicle, based at least in part, on the updated likelihood values for the plurality of candidate object trajectories for each respective object (Zeng, Para. [0068]).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to modify the processor of Chen in view of Doersch further in view of Hughes with the teachings of Zeng by including a motion planner that uses the seed trajectories to determine an optimal motion plan from the initial state to the final state. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for determining a motion plan for an autonomous vehicle or robotic system, as recognized by Zeng. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Claim 22 recites a system with elements corresponding to the processor recited in Claim 10. Therefore, the recited elements of this claim are mapped to the proposed combination in the same manner as the corresponding element in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Zeng references, presented in rejection of Claim 10, apply to this claim. Finally, the combination of the Chen, Doersch, Hughes and Zeng and references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claim 29 recites a computer-readable storage medium storing a program with instructions corresponding to the elements recited in Claim 10. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Zeng references, presented in rejection of Claim 10, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Zeng references discloses a computer readable storage medium (Doersch, Para. [0225], the computer storage medium can be a machine-readable storage device).
Claim 35 recites a robot with elements corresponding to the elements recited in Claim 10. Therefore, the elements of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Zeng references, presented in rejection of Claim 10, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Zeng references discloses one or more processors (Chen, Para. [0034], the processor 12), one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks) and a robot (Chen, Para. [0040], a mobile robot).
Claim 41 recites a method with steps corresponding to the elements of the processor recited in Claim 10. Therefore, the recited steps of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding system claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Zeng references, presented in rejection of Claim 10, apply to this claim.
Claims 11 and 23 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024) and Zeng et al. (2021/0149404 A1), as applied to claims 10, 22, 29, 35 and 41 above, and further in view of Ichnowski et al. (US 2021/0365032 A1).
Regarding claim 11, Chen in view of Doersch further in view of Hughes and Zeng teaches the processor according to claim 10,
wherein the motion planning is robot motion planning for a robot (Doersch, Para. [0222], the noisy trajectory may be conditioned on one or more actions that can be performed by the robot or mechanical agent, such that the video is a prediction of the physical environment that would be obtained if the robot or mechanical agent were to perform the one or more actions. Thus, the video may be used by a control system to control a mechanical agent such as a robot to perform a particular task by processing the predicted video using the control system to generate a control signal to control the mechanical agent, in accordance with the video, to perform the task),
wherein each seed trajectory of the plurality of seed trajectories provides a path from the initial state to the target state (Doersch, Para. [0035]; Fig. 1A, point trajectories are represented as dotted curves in FIG. 1A, with points at future time points represented as dots on the curve, and points that are closer to the tip of the curve being further in time from the input image. Point trajectories represent a dense representation of future motion of the points in the input image).
Although Chen in view of Doersch further in view of Hughes and Zeng teaches planning trajectories for a robot (Doersch, Para. [0222]), they do not explicitly teach “wherein the initial state is a joint configuration of the robot” and “wherein the target state is defined by a pose of a tool of the robot”. However, in an analogous field of endeavor, Ichnowski teaches FIG. 3 shows a set of images 300 illustrating an example motion planning trajectory 326 between an initial grasp frame 312 for a robotic arm 302 grasping an object 304 and a final grasp frame 314 of the robotic arm 302, according to an embodiment. Given the initial grasp frame 312 and the final grasp frame 314 as shown in the first image 310, the fast motion planning pipeline 200 determines the trajectory 326 of the robotic arm 302 grasping the object 304, as shown in the second image 320, such that the robotic arm 302 and the object 304 do not collide with any obstacles, such as the obstacle 316 (Ichnowski, Para. [0040]; Fig. 3).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to modify the processor of Chen in view of Doersch further in view of Hughes and Zeng with the teachings of Ichnowski by including the initial state is the joint configuration of the robot and the target state is the pose of the tool (i.e., arm) of the robot. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for planning the motion of a robot, as recognized by Ichnowski. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Claim 23 recites a system with elements corresponding to the processor recited in Claim 11. Therefore, the recited elements of this claim are mapped to the proposed combination in the same manner as the corresponding element in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes, Zeng and Ichnowski references, presented in rejection of Claim 11, apply to this claim. Finally, the combination of the Chen, Doersch, Hughes, Zeng and Ichnowski references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claims 12, 24, 30, 36 and 42 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US 2024/0013406 A1) in view of Doersch et al. (US 2024/0303897 A1, filed March 8, 2024) further in view of Hughes et al. (US 2024/0221178 A1, filed January 24, 2024), as applied to claims 1, 5, 13, 17, 25, 31 and 37 above, and further in view of Refaat et al. (US 11,586,931 B2).
Regarding claim 12, Chen in view of Doersch further in view of Hughes teaches the processor according to claim 1,
wherein the observation encoder and the noise prediction network are jointly trained via a joint training procedure (Doersch, Para. [0082], the system trains the image encoder neural network jointly with the first generative neural network), the joint training procedure comprising:
generating, based on a ground-truth trajectory and a random noise, a first training trajectory (Chen Para. [0079], the decoder is an LSTM model and generates the predicted trajectory of each target object according to the second trajectory information (i.e., representation of initial state), the third trajectory information (i.e., representation of the target state), and the noise. Para. [0073], the distribution calculator may provide different noises to the predicting model for the training of the generator network),
generating, by denoising the first trajectory based on a training environment embedding, a second training trajectory (Doersch, Para. [0094], the system processes an input for the reverse diffusion iteration that includes the noisy point trajectories as of the reverse diffusion iteration using the trajectory diffusion neural network and conditioned on the encoded representation of the input image to generate a denoising output that defines an update to the noisy point trajectories).
Although Chen in view of Doersch further in view of Hughes teaches jointly training the observation encoder and noise prediction network (Doersch, Para. [0082]), they do not explicitly teach “determining, based on the second training trajectory and the ground-truth trajectory, a loss” and “jointly updating, based on gradients of the loss, learnable weights of the observation encoder and learnable weights of the noise prediction network”. However, in an analogous field of endeavor, Refaat teaches a second loss function that evaluates a measure of difference between the training predicted future trajectory (i.e., second training trajectory) and the ground truth future trajectory (Refaat, Col. 9, lines 21-32). Refaat further teaches the system backpropagates the computed gradient of the second loss through the trajectory generation neural network into the first sub neural network to determine the update to the parameter values of the first sub neural network and the trajectory generation neural network. (Refaat, Col. 13, lines 5-9). The system can assign respective weights to the gradients of the first and second losses during different stages of the training process. A gradient that is assigned a greater weight typically results in more thorough updates to corresponding network parameter values (Refaat, Col. 13, lines 15-22).
Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention to modify the processor of Chen in view of Doersch further in view of Hughes with the teachings of Refaat by including computing a loss based on the second training trajectory and the ground-truth trajectory and updating the observation encoder and noise prediction network based on the gradients of the loss. One having ordinary skill in the art would have been motivated to combine these references because doing so would allow for training a neural network of an autonomous vehicle to make decisions based on possible trajectories, as recognized by Refaat. Thus, the claimed invention would have been obvious to one having ordinary skill in the art before the effective filing date.
Claim 24 recites a system with elements corresponding to the processor recited in Claim 12. Therefore, the recited elements of this claim are mapped to the proposed combination in the same manner as the corresponding element in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Refaat references, presented in rejection of Claim 12, apply to this claim. Finally, the combination of the Chen, Doersch, Hughes and Refaat and references discloses one or more processors (Chen, Para. [0034], the processor 12) and one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks).
Claim 30 recites a computer-readable storage medium storing a program with instructions corresponding to the elements recited in Claim 12. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Refaat references, presented in rejection of Claim 12, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Refaat references discloses a computer readable storage medium (Doersch, Para. [0225], the computer storage medium can be a machine-readable storage device).
Claim 36 recites a robot with elements corresponding to the elements recited in Claim 12. Therefore, the elements of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding processor claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Refaat references, presented in rejection of Claim 12, apply to this claim. Finally, the combination of Chen, Doersch, Hughes and Refaat references discloses one or more processors (Chen, Para. [0034], the processor 12), one or more neural networks (Chen, Para. [0038], the encoder may use other neural networks) and a robot (Chen, Para. [0040], a mobile robot).
Claim 42 recites a method with steps corresponding to the elements of the processor recited in Claim 12. Therefore, the recited steps of this claim are mapped to the proposed combination in the same manner as the corresponding elements in its corresponding system claim. Additionally, the rationale and motivation to combine the Chen, Doersch, Hughes and Refaat references, presented in rejection of Claim 12, apply to this claim.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Emma Rose Goebel whose telephone number is (703)756-5582. The examiner can normally be reached Monday - Friday 7:30-5.
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, Amandeep Saini can be reached at (571) 272-3382. 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.
/Emma Rose Goebel/Examiner, Art Unit 2662
/AMANDEEP SAINI/Supervisory Patent Examiner, Art Unit 2662
Prosecution Insights