DETAILED ACTION
This action is responsive to Applicant’s reply filed 21 April 2026. This action is made non-final.
Status of the Claims
Claims 1-2, 7-9 and 14 are amended.
Claim status is currently pending and under examination for Claims 1-5, 7-12 and 14 of which independent claims are 1 and 8.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on April 21, 2026 has been entered.
Response to Amendment
Applicant’s amendments to the Claims have overcome each and every 101 rejections previously set forth in the Final Office Action mailed January 28th 2026.
Applicant’s arguments regarding the art rejections are moot in view of the new grounds of rejection necessitated by Applicant’s amendment.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
The following are the references relied upon in the rejections below:
Sutanto, Giovanni, et al. "Learning Equality Constraints for Motion Planning on Manifolds." arXiv preprint arXiv:2009.11852 (2020).
Claims 1-2, 4, 7-9, 11 and 14 are rejected under 35 U.S.C. 102a(1) as being anticipated by Sutanto.
With respect to Claim 1, Sutanto teaches:
A method for generating training data for training a machine learning algorithm ((P. 8, Sec. 6, ¶1) “we presented a novel method called Equality Constraint Manifold Neural Network for learning equality constraint manifolds from data. … we introduced a method for augmenting a purely on-manifold dataset to include off-manifold points and several loss functions for training. This improves the robustness of the learned method while avoiding hand-coding the labels for the augmented points”),
the training data respectively comprise a data point and a data value associated with the data point ((P. 5, Sec. 4.2) “we can augment our dataset with off-manifold data to achieve a more robust learning of ECoMaNN. For each point q in the on-manifold dataset, and for each random unit vector u picked from the normal space at q, we can add an off-manifold point
q
ˇ
=
q
+
i
ϵ
u
with a positive integer i and a small positive scalar
ϵ
”),
the method comprising: providing first training data for training the machine learning algorithm ((P. 4, Sec. 4, ¶1) “One of the challenges is that the supervised training dataset is collected only from demonstrations of data points which are on the manifold
C
M
, called the on-manifold dataset.”);
approximating a manifold in which the data points of the first training data is located by ((P. 3, Sec. 3.1) “a manifold is a surface which can be well-approximated locally using an open set of a Euclidean space near every point”
(P. 8, Sec. 6, ¶1) “ECoMaNN works by aligning the row and null spaces of the local PCA and network Jacobian, which results in approximate learned normal and tangent spaces of the underlying manifold, suitable for use within a constrained sampling-based motion planner.”
(P. 4, Sec. 4, ¶2) “Using on-manifold data, the local information of the manifold can be analyzed using Local PCA”)
(i) determining, for each respective data point from the first training data, nearest neighbors of the respective data point within the data points of the first training data ((P. 4, Sec. 4, ¶2) “For each data point q in the on-manifold dataset, we establish a local neighborhood using K-nearest neighbors (KNN)”),
and (ii) forming a neighborhood graph based on the nearest neighbors of each respective data point ((P. 5, Sec. 4.2.1, ¶4) “Suppose for nearby on-manifold data points qa and qc, their approximate normal spaces … are spanned by eigenvector bases
F
N
a
… and
F
N
c
…, respectively. … during training we minimize the siamese loss … In general, the alignment of
F
N
a
and
F
N
c
is not guaranteed, for example due to the numerical sensitivity of singular value/eigen decomposition. Therefore, we introduce an algorithm for Orthogonal Subspace Alignment (OSA) in the Supplementary Material to ensure that this assumption is satisfied”
(P. 13, Sec. A.1, ¶2) “we represent the on-manifold data points as a graph. Our Orthogonal Subspace Alignment (OSA) is outlined in Algorithm 1. We begin by constructing a sparse graph of nearest neighbor connections of each on-manifold data point, followed by the construction of this graph into an (un-directed) minimum spanning tree (MST), and eventually the conversion of the MST to a directed acyclic graph (DAG).”);
determining a structure of the data points of the first training data in the manifold by, for each respective data point from the first training data, (i) performing a respective principal component analysis on the respective data point based on the nearest neighbors of the respective data point (The Examiner interprets “structure” according to its broadest reasonable interpretation in view of the applicant’s specification as encompassing K-Nearest Neighbors. This interpretation is consistent with the descriptions in the Applicant’s specification at [0013], (see excerpt below).
Applicant’s written description at [0013] “The structure of the at least one part of the data points of the first training data in the manifold is furthermore understood as a connection, or a mathematical relationship, between the coordinates of the respective data points in the manifold.”
(P. 4, Sec. 4, ¶2-3) “Using on-manifold data, the local information of the manifold can be analyzed using Local PCA. For each data point q in the on-manifold dataset, we establish a local neighborhood using K-nearest neighbors (KNN)
κ
^
=
{
q
^
1
,
q
^
2
,
…
,
q
^
K
}
, with K ≥ d. After a change of coordinates, q becomes the origin of a new local coordinate frame
F
L
, and the KNN becomes
κ
~
=
{
q
~
1
,
q
~
2
,
…
,
q
~
K
}
… Defining the matrix
X
=
[
q
~
1
q
~
2
…
q
~
K
]
T
…, we can compute the covariance matrix S … The eigendecomposition of
S
=
V
Σ
V
T
gives us the Local PCA. The matrix V contains the eigenvectors of S as its columns … These eigenvectors form the basis of the coordinate frame
F
L
. This local coordinate frame
F
L
is tightly related to the tangent space”
(P. 5, Sec. 4.2) “The training dataset is on-manifold, i.e., each point q in the dataset satisfies
h
M
q
=
0
. Through Local PCA on each of these points, we know the data-driven approximation of the normal space of the manifold at q. … For each point q in the on-manifold dataset, and for each random unit vector u picked from the normal space at q, we can add an off-manifold point
q
ˇ
=
q
+
i
ϵ
u
with a positive integer i and a small positive scalar
ϵ
(see Figure 1 for a visualization). However, if the closest on-manifold data point to an augmented point
q
ˇ
=
q
+
i
ϵ
u
is not q, we reject it. This prevents situations like augmenting a point on a sphere beyond the center of the sphere”
(P. 4, Sec. 4, Figure 1 Caption) “Figure 1: A visualization of data augmentation along the 1D normal space of a point q in 3D space. Here, purple points are the dataset, pink points are the KNN of q, and the dark red point is
q
ˇ
. q is at the axes origin, and the green plane is the approximated tangent space at that point.”
Sutanto discloses Figure 1 (reproduced below) depicting a visualization of data augmentation that uses K-Nearest Neighbors and Local Principal Component Analysis (Local PCA) to create a new augmented data point
q
ˇ
.
PNG
media_image1.png
557
698
media_image1.png
Greyscale
For each data point q in the on-manifold dataset (‘first training data’), a local neighborhood is created by using K-Nearest Neighbors. To identify the nearest neighbors of data point q, a distance metric must be used to identify which data points are most similar to point q, therefore the nearest neighbors represent a distance (mathematical) relationship with point q (and therefore determine a structure of data points of the first training data in the manifold).
Point q is then transformed to become the origin of local coordinate frame
F
L
, and the nearest neighbors of point q are also transformed. The transformed neighbors
κ
~
are used to define matrix X, and matrix X is used to compute a covariance matrix S. The eigendecomposition of covariance matrix S is performed to derive local principal components to approximate coordinate frame
F
L
(therefore performing a respective principal component analysis on the respective data point q based on the nearest neighbors of the respective data point q). To create an augmented data point
q
ˇ
(off-manifold point), point q’s nearest neighbors on coordinate frame
F
L
(derived from performing local PCA) are used.)
and (ii) determining a respective local linear approximation of the structure of the data points ((P. 2, Sec. 2.1) “Linear methods include PCA”
(P. 5, Sec. 4.2) “Through Local PCA on each of these points, we know the data-driven approximation of the normal space of the manifold at q”
Local PCA is performed for each data point q in the on-manifold dataset. Local PCA is used to approximate the local coordinate frame (‘local linear approximation’) used to find the nearest neighbors of q (see Figure 1), therefore, performing local PCA for each point q determines a local linear approximation of the structure (proximity/distance relationships) of the data points.),
the local linear approximation having coefficients derived from the respective principal component analysis of the respective data point (Since a local coordinate frame (‘local linear approximation’) is found for each data point q by using local PCA, performing the local PCA requires updating coefficients of principal components, therefore the local linear approximation has coefficients derived from the respective principal component analysis of respective data point q.);
generating additional training data based on the determined structure of the data points of the first training data in the manifold ((P. 5, Sec. 4.2) “For each point q in the on-manifold dataset, and for each random unit vector u picked from the normal space at q, we can add an off-manifold point
q
ˇ
=
q
+
i
ϵ
u
with a positive integer i and a small positive scalar
ϵ
(see Figure 1 for a visualization). However, if the closest on-manifold data point to an augmented point
q
ˇ
=
q
+
i
ϵ
u
is not q, we reject it. This prevents situations like augmenting a point on a sphere beyond the center of the sphere”
See Figure 1 depicting a new augmented data point
q
ˇ
is created by using the nearest neighbors of data point q (‘determined structure’) in an approximated local coordinate frame.);
and training, using the first training data and the additional training data, the machine learning algorithm to control at least one function of a controllable system ((P. 1, Sec. 1, ¶2) “We train ECoMaNN with datasets consisting of configurations that adhere to constraints, and present results for kinematic robot tasks learned from demonstrations.”
(P. 8, Sec. 5.3, ¶2) “The second task is a robot pick-and-place task with the additional constraint that the transported object needs to be oriented upwards throughout the whole motion. For this, we use the Orient dataset to learn the manifold for the transport phase and combine it with other manifolds that describe the pick and place operation. The planning time was 42.97s, the tree contained 1421 nodes and the optimal path had 22 nodes.”
(P. 8, Sec. 6, ¶1) “we introduced a method for augmenting a purely on-manifold dataset to include off-manifold points and several loss functions for training. This improves the robustness of the learned method while avoiding hand-coding the labels for the augmented points. We also showed that the learned manifolds can be used in a sequential motion planning framework for constrained robot tasks.”).
With respect to Claims 2 and 9, Sutanto teaches:
the method according to Claim 1, wherein: the generating the additional training data includes varying coefficients of the respective local linear approximations to determine the additional training data ((P. 5, Sec. 4.2) “Through Local PCA on each of these points, we know the data-driven approximation of the normal space of the manifold at q”
Local PCA is performed for each data point q in the on-manifold dataset. Local PCA is used to approximate the local coordinate frame (‘local linear approximation’) used to find the nearest neighbors of q (see P. 4, Sec. 4, ¶2-3). Since a local coordinate frame is found for each data point q by using local PCA, performing the local PCA requires updating coefficients of principal components, therefore each respective local coordinate frame (local linear approximation) includes varying coefficients derived from the respective principal component analysis.
See Figure 1 depicting a new augmented data point
q
ˇ
is created by using the nearest neighbors of data point q in an approximated local coordinate frame (derived from local PCA).).
With respect to Claims 4 and 11, Sutanto teaches:
the method according to Claim 1, further comprising: respectively determining, for each data point in the additional training data, a data value for the respective data point based on data values associated with the nearest neighbors of the respective data point ((P. 5, Sec. 4.2) “For each point q in the on-manifold dataset, and for each random unit vector u picked from the normal space at q, we can add an off-manifold point
q
ˇ
=
q
+
i
ϵ
u
with a positive integer i and a small positive scalar
ϵ
(see Figure 1 for a visualization). However, if the closest on-manifold data point to an augmented point
q
ˇ
=
q
+
i
ϵ
u
is not q, we reject it. This prevents situations like augmenting a point on a sphere beyond the center of the sphere”
See Figure 1 depicting a new augmented data point
q
ˇ
is created by using the nearest neighbors of data point q in an approximated local coordinate frame (derived from local PCA).).
With respect to claim 7, Sutanto teaches:
A method for controlling the at least one function of the controllable system, comprising ((P. 8, Sec. 6, ¶1) “we introduced a method for augmenting a purely on-manifold dataset to include off-manifold points and several loss functions for training. … the learned manifolds can be used in a sequential motion planning framework for constrained robot tasks):
providing a machine learning algorithm for controlling the at least one function of the controllable system, the machine learning algorithm having been trained according to the method of Claim 1 ((P. 1, Sec. 1, ¶2) “We train ECoMaNN with datasets consisting of configurations that adhere to constraints, and present results for kinematic robot tasks learned from demonstrations.”
(P. 5, Sec. 4.2) “we focus on using augmentation to aid learning an implicit constraint function for robotic motion planning”);
and controlling the at least one function of the controllable system based on the trained machine learning algorithm ((P. 8, Sec. 5.3, ¶2) “The second task is a robot pick-and-place task with the additional constraint that the transported object needs to be oriented upwards throughout the whole motion. For this, we use the Orient dataset to learn the manifold for the transport phase and combine it with other manifolds that describe the pick and place operation. The planning time was 42.97s, the tree contained 1421 nodes and the optimal path had 22 nodes.”).
With respect to Claim 8, the rejection of claim 1 is incorporated. The difference in scope being:
A control device … ((P. 8, Sec. 5.3, ¶1) “The first one is a geometric task, visualized in Figure 4, where a point starting on a paraboloid in 3D space must find a path to a goal state on another paraboloid … The task was solved in 27.09s on a 2.2 GHz Intel Core i7 processor. The tree explored 1117 nodes and the found path consists of 24 nodes.”
A path-finding task is executed on a processor, which implies a computer (‘a control device’).),
the control device comprising: a memory configured to store first training data and program code ((P. 1, Sec. 1, ¶2) “We train ECoMaNN with datasets”
Training a neural network with datasets and a processor that executes a path-finding task, implies a computer comprising a memory storing training data and program code.);
and a processor operably connected to the memory and configured to execute the program code to (A processor executing a path-finding task implies a processor operably connected to a memory and configured to execute program code.).
With respect to Claim 14, Sutanto teaches:
the control device according to Claim 8, wherein the processor further executes the program code to control the at least one function of the controllable system using the trained machine learning algorithm ((P. 8, Sec. 6, ¶1) “we introduced a method for augmenting a purely on-manifold dataset to include off-manifold points and several loss functions for training. This improves the robustness of the learned method while avoiding hand-coding the labels for the augmented points. We also showed that the learned manifolds can be used in a sequential motion planning framework for constrained robot tasks.”).
(P. 8, Sec. 5.3, ¶1) “The first one is a geometric task, visualized in Figure 4, where a point starting on a paraboloid in 3D space must find a path to a goal state on another paraboloid … The task was solved in 27.09s on a 2.2 GHz Intel Core i7 processor. The tree explored 1117 nodes and the found path consists of 24 nodes.”).
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The following are the references relied upon in the rejections below:
Villalonga, Alberto, et al. "Industrial cyber-physical system for condition-based monitoring in manufacturing processes." 2018 IEEE Industrial Cyber-Physical Systems (ICPS). IEEE, 2018.
Claims 3 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Sutanto in view of Villalonga.
With respect to Claims 3 and 10, Sutanto teaches the method according to claim 1, however, Sutanto does not teach determining nearest neighbors based on a Euclidean norm, which is taught by Villalonga:
wherein, for each data point from the first training data, the nearest neighbors are determined based on a Euclidean norm (Villalonga discloses “the procedure chosen to obtain the local model in incremental hybrid modeling is the Fuzzy k-Nearest Neighbors (F-kNN) approach. kNN consists of averaging the value of the points closest to the objective point. The kNN algorithm assumes, therefore, that nearby points have similar values. To calculate the proximity, the Euclidean norm was applied” (P. 641, Sec. 3C, First Paragraph).).
Villalonga teaches determining nearest neighbors based on a Euclidean norm is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to combine the data augmentation method of Sutanto with the technique disclosed by Villalonga to use fewer computing resources. Calculating a Euclidean norm is computationally efficient because a Euclidean norm relies on simple arithmetic operations (addition, squaring, square root) to calculate a distance. Computers are equipped with processors that are optimized to handle simple arithmetic operations, therefore these calculations would not take a long time to compute or use many computational resources.
The following are the references relied upon in the rejections below:
Javed, Abbas, et al. "Design and implementation of a cloud enabled random neural network-based decentralized smart controller with intelligent sensor nodes for HVAC." IEEE Internet of Things Journal 4.2 (2016): 393-403.
Claims 5 and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Sutanto in view of Javed.
With respect to Claims 5 and 12, Sutanto teaches the method according to Claim 1, however, Sutanto does not teach using training data comprised of sensor data, which is taught by Javed:
wherein the first training data comprise sensor data (Javed discloses “the trained RNN model is implemented on an indoor environment sensor node. The inputs for the model are: HVAC inlet air temperature, HVAC inlet air CO2 concentrations, inlet air temperature of the environment chamber, and CO2 concentration inside the environment chamber. The output of the model is the number of occupants inside the environment chamber. The training data set for the RNN model is downloaded from the Web portal” (P. 397, Sec. 4B, First Paragraph).
Javed further discloses “for each sensor node, the Web portal displays node ID, upload time in milli seconds (the time sensor node is powered), light intensity, CO2 concentrations, temperature, humidity, dewpoint temperature, data receiving time, motion sensor, heating setpoint, cooling setpoint, heating output for HVAC, cooling output for HVAC, ventilation output for HVAC, and number of occupants in the room” (P. 397, Sec. 3F).).
Javed teaches using training data gathered from sensors to train a machine learning model is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to combine the data augmentation method of Sutanto with the technique disclosed by Javed to train an accurate machine learning model. Sensors provide accurate, specific, and real-time data about an environment, which can be used to train a machine learning model that captures contextualized relationships and patterns. By using a trained model that has learned complex patterns and relationships, the model can predict accurate outcomes, which can be used to make well-informed decisions.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PEDRO J MORALES whose telephone number is (571)272-6106. The examiner can normally be reached 8:30 AM - 6:00 PM.
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, MIRANDA M HUANG can be reached at (571)270-7092. 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.
/PEDRO J MORALES/Examiner, Art Unit 2124
/VINCENT GONZALES/Primary Examiner, Art Unit 2124
Prosecution Insights