DETAILED ACTION
Examiner Remarks
In light of Applicant’s Remarks and Amendments submitted on 12/11/2025, Examiner has withdrawn the objections to the claims and the 112(b) rejections. The objection to the Specification has not been withdrawn since Applicant’s Remarks submitted on 12/11/2025 did not address nor fix the informalities found in the Specification as detailed by the Non-Final Rejection of 06/13/2025.
Response to Arguments
Applicant argues that there was no rationale to combine the prior art references of Zhang and Shih. See pgs., 10-11 of Applicant’s Remarks submitted on 12/11/2025.
Respectfully, Examiner disagrees. As a preliminary matter, references were not just “spliced” together but rather a combination was formed to reject Applicant’s claims. Unlike what Applicant states in page 11, Examiner did not justify the prior art combination based upon Applicant’s Specification but rather the motivation to combine (as once again stated in the current Office-Action) dealt with devising a prediction system to avoid collusions among vehicles, which the prior art of Shih teaches. See Current Office Action for the detailed teaching. This type of knowledge would have been available to a person of ordinary skill in the art at the time of the effective filing date of Applicant’s invention. Again, it must be recognized and restated that any judgment on obviousness is in a sense necessarily a reconstruction based upon hindsight reasoning. But so long as it takes into account only knowledge which was within the level of ordinary skill at the time the claimed invention was made, and does not include knowledge gleaned only from the applicant's disclosure, such a reconstruction is proper. See In re McLaughlin, 443 F.2d 1392, 170 USPQ 209 (CCPA 1971)(Emphasis added). Accordingly, the 103 rejection has not been withdrawn since impermissible hindsight was not used.
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
Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. EP 20155183.5, filed on 02/03/2020.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 02/17/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Specification
The disclosure is objected to because of the following informalities:
Pg., 12, lines 3-16, do not have the corresponding fig. numbers for components 102a and 102b, since fig. 2 does not contain components 102a and 102b
Appropriate correction is required.
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 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, 3-4, 9-10, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang, Yuxuan, et al. "Trafficgan: Network-scale deep traffic prediction with generative adversarial nets." IEEE Transactions on Intelligent Transportation Systems 22.1 (2019) (“Zhang”) in view of Goodfellow, Ian, et al. "Generative adversarial nets." Advances in neural information processing systems 27 (2014)(“Goodfellow”) and in view of Dieter et al., “Modellbildung und Simulation der Dynamik von Kraftfahrzeugen,” springer, Berlin/Heidelberg, 2010(“Dietrich”) and further in view of Shih, Chi-Sheng, et al. "Vehicle speed prediction with RNN and attention model under multiple scenarios." 2019 IEEE Intelligent Transportation Systems Conference (ITSC). IEEE, 2019(“Shih”).
Regarding claim 1, Zhang teaches a method for determining a velocity of a vehicle, comprising the following steps:
providing an input for a first generative model depending on a route information, [a probabilistic variable including noise, and depending on an output of a second physical model] (Zhang, pg., 3, see also fig. 1 and 3, “We use a traffic network matrix
S
t
to characterize the traffic conditions of a transportation network G in the time slot t. The rows and columns of the matrix correspond to the intersections V of G, and each element
s
i
j
t
of the matrix is the traffic flow on the road segment between two intersections… [c]onsidering the significant advantages of the GAN model in image generation and video prediction, it motivates us to use it for traffic flow prediction on a transportation network.” & Zhang, pg., 5, “[The training mechanism of GAN is as follows]
min
g
max
D
f
D
,
G
=
E
S
t
n
+
1
~
P
d
a
t
a
S
t
n
+
1
l
o
g
D
S
t
n
+
1
+
E
z
~
P
F
(
F
)
[
log
1
-
D
G
F
]
.” Zhang teaches: fig. 3 which details inputting a traffic network matrix into a generator/discriminator architecture(i.e. GAN) where GAN samples the probabilistic value Z from prior distribution of
P
F
(
F
)
in the following minmax equation:
min
g
max
D
f
D
,
G
=
E
S
t
n
+
1
~
P
d
a
t
a
S
t
n
+
1
l
o
g
D
S
t
n
+
1
+
E
z
~
P
F
(
F
)
[
log
1
-
D
G
F
]
which maps to: providing an input for a first generative model depending on a route information);1
and determining an output of the first generative model in response to the input for the first generative model, wherein the output of the first generative model characterizes the velocity(Zhang, pg., 3, see also fig. 1, “The construction of a traffic network matrix is illustrated in Fig. 1. One can see that there is a road segment
l
5,8
between intersections inter5 and inter8. The traffic flow on the road
segment
l
5,8
and
l
8,5
are 16 mph and 13 mph, respectively…[i]n this way, the traffic conditions on all the road segments of a transportation network in the time slot t are modeled as a traffic network matrix
S
t
…we predict the traffic network matrix
S
t
n
+
1
.” Zhang teaches: The construction of a traffic network matrix is illustrated in Fig. 1. The traffic flow on the road
segment
l
5,8
and
l
8,5
are 16 mph and 13 mph where the transportation network in the time slot t are modeled as a traffic network matrix and we predict the traffic network matrix which maps to: and determining an output of the first model in response to the input for the first model, wherein the output of the first model characterizes the velocity);
wherein the first generative model includes a first component trained to map the input for the first generative model determined depending on the route information [and the probabilistic variable] to an intermediate output which is an acceleration(Zhang, pgs. 7-9, see also fig. 11, “In order to evaluate the effectiveness of TrafficGAN, we first define the traffic conditions of the road segments based on the average speed of traffic flow. Average speeds of 0-9, 10-14, 15-19, 20-24, and over 25 mph correspond to the traffic conditions of heavy congestion, medium-heavy congestion, medium congestion, light congestion and free flow for the road segments, respectively[wherein the first model includes a first component trained to map the input for the first model determined depending on the route information]…[i]n order to more clearly display the performance of Traffic-GAN, we compare the prediction results of several preferable models on a same segment at different times of a day. Fig 11 presents the predicted average speeds for TrafficGAN[to an intermediate output which is an acceleration]…as well as the real data. One can see that the prediction of TrafficGAN (red curve) better fits with the curve of the real data (blue curve) and more accurately reflects the variation trend of the average speed.”)2
[and determining a characteristic of the velocity over time depending on]a plurality of inputs for the first generative model and [a plurality of inputs for the second model, wherein a series of values for the velocity is determined as the characteristic of the velocity over time](Zhang, pgs. 7-9, see also fig. 11, “The dataset used for evaluation are collected from Chicago Transit Authority (CTA) buses on Chicago’s arterial streets (non-freeway streets) in real-time by continuously monitoring and analyzing their GPS traces…[i]n order to evaluate the effectiveness of TrafficGAN, we first define the traffic conditions of the road segments based on the average speed of traffic flow[a plurality of inputs for the first generative model and].”).3
While Zhang does teach a first generative model and the first generative model, Zhang does not teach: a probabilistic variable including noise; and the probabilistic variable.
However, Goodfellow teaches
[providing an input for a first generative model depending on a route information], a probabilistic variable including noise, [and depending on an output of a second physical
Model] (Goodfellow, pg. 2, “To learn the generator’s distribution
p
g
over data
x
, we define a prior on input noise variables
p
z
(
z
)
then represent a mapping to data space as G(z;
θ
g
) [where G represent the generator and z represents the sampled noise from the noise distribution
p
z
(
z
)
]…”);4
[wherein the first generative model includes a first component trained to map the input for the first generative model determined depending on the route information] and the probabilistic variable [to an intermediate output which is an acceleration](Goodfellow, pg. 2, “To learn the generator’s distribution
p
g
over data
x
, we define a prior on input noise variables
p
z
(
z
)
then represent a mapping to data space as G(z;
θ
g
) [where G represent the generator and z represents the sampled noise from the noise distribution
p
z
(
z
)
]…”).5
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang with the teachings of Goodfellow the motivation to do so would be to add random noise to the generator to avoid using approximate inference or Markov chains during training(Goodfellow, pg. 2, “[W]e explore the special case when the generative model generates samples by passing random noise through a multilayer perceptron…[w]e refer to this special case as adversarial nets. In this case, we can train both models using only the highly successful backpropagation and dropout algorithms…and sample from the generative model using only forward propagation…[with] [n]o approximate inference or Markov chains…[being] necessary.”).
While Zhang in view of Goodfellow do teach a first generative model, the first generative model, and the probabilistic variable including noise, Zhang in view of Goodfellow do not teach: and depending on an output of a second physical model.
However, Dietrich teaches:
[providing an input for a first generative model depending on a route information, a probabilistic variable including noise,] and depending on an output of a second physical model(Dietrich, 292, see also fig. 11.8 (as reproduced herein), “The vehicle model described below supplements the model described in the previous section with wheel suspension kinematics.”
PNG
media_image1.png
277
600
media_image1.png
Greyscale
);6, 7
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow with the teachings of Dietrich the motivation to do so would be to create a realistic model to better incorporate the dynamics of a real vehicle for different driving patterns(Dietrich, pgs. 8, “The design and testing of such systems with their enormous variety of functions is a high challenge. Requirements for design methods and testing programs and the resulting modeling and simulation technology... driving maneuvers can be simulated as often and reproducibly as desired under defined boundary conditions... critical driving maneuvers can be replaced by safe simulation.”).8
While Zhang in view of Goodfellow and Dietrich do teach a first generative model, a second physical model, the first generative model, and the probabilistic variable including noise, Zhang in view of Goodfellow and Dietrich do not teach: includes a second component trained to map the intermediate output to the velocity of the vehicle depending on an output of a second
model; wherein the output of the second model characterizes a physical constraint for the
intermediate output, wherein the physical constraint is based on the route information;
providing an input for the second model depending on at least one vehicle state or
the route information; determining the output of the second model in response to the input for the second model; and determining a characteristic of the velocity over time depending on a plurality of inputs for the second model, wherein a series of values for the velocity is determined as the characteristic of the velocity over time
However, Shih teaches:
[wherein the first generative model] includes a second component trained to map the
intermediate output to the velocity of the vehicle depending on an output of a second
model(Shih, pgs., 371-374, see also figs. 6 and 9, “Figure 9 illustrates the flow of the proposed EDA network, which consists of four major steps… [t]he model takes a sequence of velocity. Each velocity, V , is fed into a uniform quantization with 256 classes… [a]fter the quantization step, one-hot encoding is conducted so as to take a sequence of velocity and generate a prediction for 5 seconds… there are 50 points for 5 seconds for each input sequence, shown in Eqn. 2, and output sequence, shown in Eqn. 3[includes a second component trained to map the
intermediate output to the velocity of the vehicle depending on an output of a second
model]… [a]t the end of the flow, we need to do inverse quantization for calculation of MSE.”);9
wherein the output of the second model characterizes a physical constraint for the
intermediate output, wherein the physical constraint is based on the route information(Shih, pgs., 371-374, see also figs. 6 and 9, “The proposed EDA networks[wherein the output of the second model characterizes], shown in Figure 6, starts with data classifier to determine the driving scenario and continues to feed the aligned input sequence into prediction networks… [t]he data classifier annotates the input sequences for different driving scenarios so that the subsequent steps can use different models to predict the vehicle velocity… [for] [t]he second classifier annotates the speed limit at the vehicle location, which are available from navigation map services. In this work, the algorithm query openstreetmap to obtain the road types and speed limit based on the vehicle location[a physical constraint for the
intermediate output, wherein the physical constraint is based on the route information].”);
providing an input for the second model depending on at least one vehicle state or
the route information(Shih, pgs., 371-374, see also figs. 6 and 9, “The proposed EDA networks, shown in Figure 6, starts with data classifier to determine the driving scenario and continues to feed the aligned input sequence into prediction networks…[for] [t]he second classifier annotates the speed limit at the vehicle location, which are available from navigation map services. In this work, the algorithm query openstreetmap to obtain the road types and speed limit based on the vehicle location. For each road segment of the trajectory, the algorithm calculates the average speed of the vehicle and annotates it into five classes. When compared with the speed limit, the average speed class will serve as an indication of driver intention and traffic condition[providing an input for the second model depending on at least one vehicle state or the route information].”);
determining the output of the second model in response to the input for the second
model; and determining a characteristic of the velocity over time depending on [a plurality of inputs for the first generative model and] a plurality of inputs for the second model, wherein a series of values for the velocity is determined as the characteristic of the velocity over time(Shih, pgs., 373-375, see also Table III, “Two datasets are used for the sake of generosity… [t]he first dataset is a private dataset collected from nine bus routes… [t]he collected data include: value from gyroscope in x, y, z directions at 10Hz, acceleration in x,y,z direction at 10Hz, latitude and longitude coordinates at 0.5Hz,GPS speed at 0.5Hz, wheel speed at 2Hz… [t]he time span of each route ranges from 2min to 20min… [h]ence, by detecting the change of lane ID, we could annotate the lane change. Last, the data in NGSIM dataset are all collected on highway… [i]n the experiment, the three maneuver classes, including change lane, go straight, and turn, are compared individually…[t]able III shows the predication errors in private dataset while EDA network model is used[depending on a plurality of inputs for the second model; determining the output of the second model in response to the input for the second model]. The results show that the
PNG
media_image2.png
171
530
media_image2.png
Greyscale
model does have different performance for different driving behaviors. When the vehicle always drive straight, the model predicts best; when the vehicle changes lane or makes turn, the model leads to higher predication errors although the difference ranges from 0.09 meter per second to 0.11 meter per second[and determining a characteristic of the velocity over time, wherein a series of values for the velocity is determined as the characteristic of the velocity over time]”).10
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow and Dietrich with the teachings of Shih the motivation to do so would be to devise a prediction system to avoid collusions among vehicles(Shih, pg., 369, “Free space defines the geometry region for a vehicle to safely move in a certain time period and is essential for every vehicle, controlled by both human and computer software, to prevent collision. Over-conservative estimation for free space will not only reduce the utilization of the shared space but also lead to traffic accident. Motion prediction of nearby vehicles are essential to estimate free space and is the goal of this work.”).
Regarding claim 3, Zhang in view of Goodfellow, Dietrich, and Shih teaches the method according to claim 1, wherein the physical constraint for a time step is determined depending on the velocity of the vehicle in a previous time step, and/or a force applied to the vehicle and/or a force applied by the vehicle(Dietrich, pg., 311, sec. 11.2.7, “The force application points of the body spring and the body damper should still be freely definable when parameterizing the model. For this reason, different points of application of the force elements are defined for the spring and the damper”).11,12
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, and Shih with the above teachings of Dietrich for the same rationale stated at Claim 1.
Regarding claim 4, Zhang in view of Goodfellow, Dietrich, and Shih teaches the method according to claim 1, wherein the route information includes at least one of: (i) a geographical characteristic including an absolute height or a road slope characteristic, (ii) a traffic flow characteristic including a time dependent average speed of traffic, (iii) a road characteristic including a number of lanes, and/or road type and/or road curvature, (iv) a traffic control characteristic including a speed limit characteristic, and/or a number of traffic lights, and/or a number of traffic signs of a specific type, and/or a number of stop signs, and/or a number of yield signs, and/or a number of pedestrian crossing signs, (v) a weather characteristic including an amount of rain at a predetermined time, and/or a wind speed, and/or a presence of fog(Zhang, pg. 7, “The dataset used for evaluation are collected from Chicago Transit Authority (CTA) buses on Chicago’s arterial streets (non-freeway streets) in real-time by continuously monitoring and analyzing their GPS traces… [t]he average speed data of the CTA buses are produced every 10 minutes on the road segments divided through the intersections… [t]here are usually two traffic flows with the opposite directions on one road (except for the one-way street), and one road they have different road segment IDs. The dataset includes the following information: Time, segmentID (ID number of each road segment), BusCount (how many buses running on this road segment currently), ReadCount (the number of readings sent by the buses), and Speed (the average speed of the buses). We also have the following information about each road segment: street name, direction, start street name, end street name, length, the start and end locations of the segment.”).13
Regarding claim 9, Zhang in view of Goodfellow, Dietrich, and Shih teaches the method according to claim 1, wherein a start velocity is determined, and wherein a succeeding velocity is determined depending on the start velocity(Dietrich, pgs. 299-300, “Based on Eqs. (11.80) to (11.83) the absolute velocities can be given as:
v
T
i
=
r
˙
T
i
=
r
˙
v
+
ω
v
×
r
V
T
i
+
r
˙
v
T
i
.” Dietrich teaches:
r
V
T
i
which represents the start velocity and
v
T
i
represents the succeeding velocity based on the start velocity
r
V
T
i
which maps to: wherein a start velocity is determined, and wherein a succeeding velocity is determined depending on the start velocity).14
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, and Shih with the above teachings of Dietrich for the same rationale stated at Claim 1.
Regarding claim 10, Zhang in view of Goodfellow, Dietrich, and Shih teaches the method according to claim 9, wherein the start velocity is either set to zero or wherein the start velocity is determined as output of a start-velocity-model including an artificial neural network trained to map the route information to the start velocity(Zhang, pg. 7, “In order to evaluate the effectiveness of TrafficGAN, we first define the traffic conditions of the road segments based on the average speed of traffic flow. Average speeds of 0-9…correspond to the traffic conditions of heavy congestion….” & Zhang, pg. 8, see also fig. 8, “The output from each LSTM cell in D is fed into a fully connected layer with shared weights across the time steps, and then a sigmoid function is used for the final prediction… Fig. 8 compares the average speed of traffic flows of all the road segments at weekdays and weekends.” Zhang teaches: the Average speeds of range from 0 to 9 which maps to: wherein the start velocity is either set to zero; Zhang teaches: The output from each LSTM cell in D is fed into a fully connected layer and then a sigmoid function is used for the final prediction to output the average speed of traffic flows as detailed by fig. 8 which maps to: or wherein the start velocity is determined as output of a start-velocity-model including an artificial neural network trained to map the route information to the start velocity).15
Referring to independent claim 15 Goodfellow teaches a non-transitory computer-readable storage medium on which is stored a computer program including computer readable instructions(Goodfellow, pg. 5, “We trained adversarial nets [on]…a range of datasets including MNIST… [and] the Toronto Face Database….”)16 and for all other claim limitations of claim 15 they are rejected on the same basis as independent claim 1 since they are analogous claims.
Claims 5, 7, 11-12 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang, Yuxuan, et al. "Trafficgan: Network-scale deep traffic prediction with generative adversarial nets." IEEE Transactions on Intelligent Transportation Systems 22.1 (2019) (“Zhang”) in view of Goodfellow, Ian, et al. "Generative adversarial nets." Advances in neural information processing systems 27 (2014)(“Goodfellow”) and in view of Dieter et al., “Modellbildung und Simulation der Dynamik von Kraftfahrzeugen,” springer, Berlin/Heidelberg, 2010(“Dietrich”) and in view of Shih, Chi-Sheng, et al. "Vehicle speed prediction with RNN and attention model under multiple scenarios." 2019 IEEE Intelligent Transportation Systems Conference (ITSC). IEEE, 2019(“Shih”) and further in view of Yufang et al., "Investigating long‐term vehicle speed prediction based on BP‐LSTM algorithms." IET Intelligent Transport Systems 13.8 (2019) (“Yufang”).
Regarding claim 5, Zhang in view of Goodfellow, Dietrich, and Shih teaches the method according to claim 1, but does not teach: further comprising: providing an input for a third model depending on the route information and the velocity; and determining an output of the third model in response to the input for the third model; wherein the output of the third model characterizes a score indicating an estimate of veracity for the velocity; wherein the third model is trained to map the input for the third model determined depending on the route information and the velocity to output of the third model characterizing the score indicating the estimate of veracity for the velocity.
However, Yufang teaches further comprising:
providing an input for a third model depending on the route information and the velocity(Yufang, pgs. 1282-1283, see also fig. 4, “A sedan car with 2.0 L ICE and a 6-speed AT is selected as the data collection vehicle with an unchanged driver. Research ideas are shown in Fig. 4…driving data are collected and recorded, including time (t), actual vehicle speed (v), single travel distance (dis), acceleration (a), relative vehicle speed of the front vehicle (dv), relative distance from the front vehicle (dx), number of stops (stops), position information such as longitude (Lng) and latitude (Lat).Through application transportation interface (API), GIS and
traffic flow state, including road type (road), speed limit, and traffic congestion degree… will be collected and stored in the local server for the establishment of prediction model.” Yufang teaches: driving data are collected and recorded, including time (t), actual vehicle speed (v) position information such as longitude (Lng) and latitude (Lat) for the establishment of prediction model which maps to: providing an input for a third model depending on the route information and the velocity);
and determining an output of the third model in response to the input for the third model(Yufang, pgs. 1285-1286, see also figs. 7 and 8, “The long-term vehicle speed prediction algorithm designed in this paper is characteri[z]ed by the ability to select different suitable algorithms to predict the speed of the vehicle according to the type of road segment, involving BP algorithm prediction on Known/Unknown city roads, LSTM algorithm prediction on Known/Unknown suburb roads and freeway roads. Among them, the predicted vehicle speed can basically keep up with the change law of the real vehicle speed, as is shown in Figs. 7 and 8….” Yufang teaches: The long-term vehicle speed prediction algorithm designed in this paper is characterized by the ability to select different suitable algorithms to predict the speed of the vehicle according to the type of road segment which maps to: and determining an output of the third model in response to the input for the third model );
wherein the output of the third model characterizes a score indicating an estimate of veracity for the velocity(Yufang, pgs. 1285-1286, see also fig. 9,“ According to the BP-LSTM prediction algorithm, the prediction deviation is shown in Fig. 9. From Fig. 9a, the prediction deviation on city road is mostly within 5…[f]rom Fig. 9b, the prediction results on suburb road with Known path are mostly concentrated within 3.2… [f]rom Fig. 9c, the prediction deviation on the Known freeway path are mostly concentrated near 4 and mostly near 9 on the Unknown freeway path….” Yufang teaches: According to the BP-LSTM prediction algorithm, the prediction deviation is shown in Fig. 9. From Fig. 9a, the prediction deviation on city road is mostly within 5 while on suburb road with Known path are mostly concentrated within 3.2 which maps to: wherein the output of the third model characterizes a score indicating an estimate of veracity for the velocity );
wherein the third model is trained to map the input for the third model determined depending on the route information and the velocity to output of the third model characterizing the score indicating the estimate of veracity for the velocity(Yufang, pgs. 1282-1283, see also fig. 4, “A sedan car with 2.0 L ICE and a 6-speed AT is selected as the data collection vehicle with an unchanged driver. Research ideas are shown in Fig. 4…driving data are collected and recorded, including time (t), actual vehicle speed (v), single travel distance (dis), acceleration (a), relative vehicle speed of the front vehicle (dv), relative distance from the front vehicle (dx), number of stops (stops), position information such as longitude (Lng) and latitude (Lat) … will be collected and stored in the local server for the establishment of prediction model.” & Yufang, pgs. 1285-1286, see also fig. 9,“ According to the BP-LSTM prediction algorithm, the prediction deviation is shown in Fig. 9. From Fig. 9a, the prediction deviation on city road is mostly within 5…[f]rom Fig. 9b, the prediction results on suburb road with Known path are mostly concentrated within 3.2… [f]rom Fig. 9c, the prediction deviation on the Known freeway path are mostly concentrated near 4 and mostly near 9 on the Unknown freeway path….” & Yufang, pg. 1284, section 4.1, “[For the BP Model] [i]t is composed of two
hidden layers. The specific parameter settings are shown in Table 2 [reproduced herein]…[and for the LSTM] A four-layer LSTM model is built, which is shown in Table 3[reproduced herein].”
PNG
media_image3.png
268
511
media_image3.png
Greyscale
PNG
media_image4.png
206
469
media_image4.png
Greyscale
As Tables 2 and 3 detail the BP portion of the model uses the training function of trainlm and trains for 100 steps, using a step size of 0.02 and the LSTM portion of the model uses the Adam optimizer to train for 100 epochs with a learning rate of 0.001 and batch size of 1734).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, Dietrich and Shih with the teachings of Yufang the motivation to do so would be to predict a vehicle’s speed using a hybrid neural network that is able to learn the underlying behavior of drivers for better time prediction and energy consumption(Yufang, pg. 1281, “[B]y comparing the efficiency and rationality of the several algorithms, a novel data-driven back propagation-long short-term memory (BP-LSTM) algorithm for on-road individual long-term average vehicle speed prediction along a driving route is proposed…[t]he novelty of the proposed method lies in the consideration of unobservable driving states in the prediction model hidden in the collected individual driving data, such as the driver's behavio[]rs… the long-term vehicle speed prediction result in this paper can be used well for travel time prediction and energy consumption prediction scenarios, which can reali[z]e the intelligent navigation and energy saving for new energy vehicles.”).
Regarding claim 7, Zhang in view of Goodfellow, Dietrich, Shih and Yufang teaches the method according to claim 5, further comprising: providing the route information as a continuous or discrete first series of values over time within a time period(Zhang, pg., 3-5, see also fig. 1 and 3, “We use a traffic network matrix
S
t
to characterize the traffic conditions of a transportation network G in the time slot t. The rows and columns of the matrix correspond to the intersections V of G, and each element
s
i
j
t
of the matrix is the traffic flow on the road segment between two intersections….”);
providing the probabilistic variable as a continuous or discrete second series of values over time within the time period(Zhang,, pgs. 4-5, “The training mechanism of GAN is as follows…where
P
d
a
t
a
(
S
t
n
+
1
)
is the real data distribution and
P
F
(
F
)
is the prior distribution …
min
g
max
D
f
D
,
G
=
E
S
t
n
+
1
~
P
d
a
t
a
S
t
n
+
1
l
o
g
D
S
t
n
+
1
+
E
z
~
P
F
(
F
)
[
log
1
-
D
G
F
]
”));
determining, by the first generative model, a continuous or discrete third series of values for the characteristic of velocity over time depending on the values of the first series and the second series(Zhang, pg., 3, see also figs. 1 and 2, “The construction of a traffic network matrix is illustrated in Fig. 1. One can see that there is a road segment
l
5,8
between intersections inter5 and inter8. The traffic flow on the road segment
l
5,8
and
l
8,5
are 16 mph and 13 mph, respectively…[i]n this way, the traffic conditions on all the road segments of a transportation network in the time slot t are modeled as a traffic network matrix
S
t
…we predict the traffic network matrix
S
t
n
+
1
);
and determining, by the third model, the score depending on the values of the first series and the third series(Yufang, pgs. 1282-1283, see also fig. 4, “A sedan car with 2.0 L ICE and a 6-speed AT is selected as the data collection vehicle with an unchanged driver. Research ideas are shown in Fig. 4…driving data are collected and recorded, including time (t), actual vehicle speed (v), single travel distance (dis), acceleration (a), relative vehicle speed of the front vehicle (dv), relative distance from the front vehicle (dx), number of stops (stops), position information such as longitude (Lng) and latitude (Lat).Through application transportation interface (API), GIS and traffic flow state….” & Yufang, pgs. 1285-1286, see also fig. 9,“ According to the BP-LSTM prediction algorithm, the prediction deviation is shown in Fig. 9. From Fig. 9a, the prediction deviation on city road is mostly within 5…[f]rom Fig. 9b, the prediction results on suburb road with Known path are mostly concentrated within 3.2… [f]rom Fig. 9c, the prediction deviation on the Known freeway path are mostly concentrated near 4 and mostly near 9 on the Unknown freeway path….” Yufang teaches: According to the BP-LSTM prediction algorithm, the prediction deviation is shown in Fig. 9. From Fig. 9a, the prediction deviation on city road is mostly within 5 while on suburb road with Known path are mostly concentrated within 3.2 which maps to: determining, by the third model, the score);
wherein the first generative model is a first Recurrent Neural network(Zhang, pg., 4, see also figs. 3, “As shown in Fig. 3, the generator contains three layers to capture the spatial-temporal features of the input traffic data. The observed traffic matrix sequence is first input a Convolution Neural Network (CNN) layer to learn the spatial features of the traffic data on the entire road network. Then a Long-Short Term Memory (LSTM) layer is used to capture
the temporal correlations of the sequential traffic data. The output of the LSTM layer is next input into another CNN layer to generate the new traffic network matrices in the
future time slots. The discriminator part also contains three layers, the CNN layer, the Bi-directional LSTM layer and the fully connected layer.”);
and wherein the third model is a second Recurrent Neural network(Yufang, pg., 1282, see also fig. 3, “The structure of the LSTM algorithm is shown in Fig. 3. Fig. 3 shows the vehicle speed prediction scheme based on LSTM, and the inputs gates are set to control inputs information….”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, Dietrich, and Shih with the above teachings of Yufang for the same rationale stated at Claim 5.
Regarding claim 11, Zhang in view of Goodfellow, Dietrich, Shih and Yufang teaches the method according to claim 5, wherein the velocity is determined depending on the output of the first generative model and the second model in response to training data defining input data for the first generative model(Zhang, pg., 3-5, see also fig. 1 and 3, “We use a traffic network matrix
S
t
to characterize the traffic conditions of a transportation network G in the time slot t. The rows and columns of the matrix correspond to the intersections V of G, and each element
s
i
j
t
of the matrix is the traffic flow on the road segment between two intersections….”) and the second model(Dietrich, pgs. 322-323, see also table 11.8, “In Table 11.8 some input values for the vehicle models described are given as examples.”),17
wherein the output of the third model characterizing the score indicating the estimate of veracity for the velocity is determined, and wherein at least one parameter of the first generative model and/or the second model and/or the third model is determined depending on the score(Yufang, pg. 1286, “According to the BP-LSTM prediction algorithm, the prediction deviation is shown in Fig. 9. From Fig. 9a, the prediction deviation on city road is mostly within 5, and the overall prediction deviation shows a decreasing convergence trend with the increase of the prediction absolute difference value. From Fig. 9b, the prediction results on suburb road with Known path are mostly concentrated within 3.2, and the overall prediction deviation shows a decreasing convergence trend with the increase of the prediction absolute difference value….”).18
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, Dietrich, and Shih with the above teachings of Yufang for the same rationale stated at Claim 5.
Regarding claim 12, Zhang in view of Goodfellow, Dietrich, Shih and Yufang teaches the method according to claim 11, further comprising providing input data including the velocity, the route information, the intermediate output, and the at least one vehicle state(Yufang, pgs. 1282-1283, see also figs. 4, 5 and table 1, “With the assistance of controller area network (CAN) interface and sensors, driving data are collected and recorded, including time (t), actual vehicle speed (v), single travel distance (dis), acceleration (a), relative vehicle speed of the front vehicle (dv), relative distance from the front vehicle (dx), number of stops (stops), position information such as longitude (Lng) and latitude (Lat).”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, Dietrich, and Shih with the above teachings of Yufang for the same rationale stated at Claim 5.
Regarding claim 14, Zhang teaches a non-transitory machine learning system, comprising
a first generative model(Zhang, pg., 4, see also fig. 3, “As shown in Fig. 3, the generator contains three layers to capture the spatial-temporal features of the input traffic
data. The observed traffic matrix sequence is first input a Convolution Neural Network (CNN) layer to learn the spatial features of the traffic data on the entire road network. Then a Long-Short Term Memory (LSTM) layer is used to capture the temporal correlations of the sequential traffic data. The output of the LSTM layer is next input into another CNN layer to generate the new traffic network matrices in the future time slots. The discriminator part also contains three
layers, the CNN layer, the Bi-directional LSTM layer and the fully connected layer. The generated and real traffic network matrices are first input to the CNN layer to learn the latent
spatial features, and then input to a Bi-directional LSTM layer to capture the latent temporal features. Next a fully connected layer is used to transform the output of the Bidirectional
LSTM to a low-dimensional feature vector. Finally, a classification model is built with the feature vectors to classify whether the input future traffic network matrix is real or fake.”);
the non-transitory machine learning system configured to determine a velocity of a vehicle(Zhang, pg., 3, see also fig. 1, “The construction of a traffic network matrix is illustrated in Fig. 1. One can see that there is a road segment
l
5,8
between intersections inter5 and inter8. The traffic flow on the road segment
l
5,8
and
l
8,5
are 16 mph and 13 mph, respectively…[i]n this way, the traffic conditions on all the road segments of a transportation network in the time slot t are modeled as a traffic network matrix
S
t
…we predict the traffic network matrix
S
t
n
+
1
.” Zhang teaches: The construction of a traffic network matrix is illustrated in Fig. 1. The traffic flow on the road
segment
l
5,8
and
l
8,5
are 16 mph and 13 mph where the transportation network in the time slot t are modeled as a traffic network matrix and we predict the traffic network matrix which maps to: the machine learning system configured to determine a velocity of a vehicle);
the non-transitory machine learning system configured to:
provide an input for a first generative model depending on a route information,[a probabilistic variable including noise, and depending on an output of the second physical model](Zhang, pg., 3, see also fig. 1 and 3, “We use a traffic network matrix
S
t
to characterize the traffic conditions of a transportation network G in the time slot t. The rows and columns of the matrix correspond to the intersections V of G, and each element
s
i
j
t
of the matrix is the traffic flow on the road segment between two intersections… [c]onsidering the significant advantages of the GAN model in image generation and video prediction, it motivates us to use it for traffic flow prediction on a transportation network.” & Zhang, pg., 5, “[The training mechanism of GAN is as follows]
min
g
max
D
f
D
,
G
=
E
S
t
n
+
1
~
P
d
a
t
a
S
t
n
+
1
l
o
g
D
S
t
n
+
1
+
E
z
~
P
F
(
F
)
[
log
1
-
D
G
F
]
.” Zhang teaches: fig. 3 which details inputting a traffic network matrix into a generator/discriminator architecture(i.e. GAN) where GAN samples the probabilistic value Z from prior distribution of
P
F
(
F
)
in the following minmax equation:
min
g
max
D
f
D
,
G
=
E
S
t
n
+
1
~
P
d
a
t
a
S
t
n
+
1
l
o
g
D
S
t
n
+
1
+
E
z
~
P
F
(
F
)
[
log
1
-
D
G
F
]
which maps to: provide an input for a first generative model depending on a route information);19
and determine an output of the first generative model in response to the input for the first generative model, wherein the output of the first generative model characterizes the velocity(Zhang, pg., 3, see also fig. 1, “The construction of a traffic network matrix is illustrated in Fig. 1. One can see that there is a road segment
l
5,8
between intersections inter5 and inter8. The traffic flow on the road
segment
l
5,8
and
l
8,5
are 16 mph and 13 mph, respectively…[i]n this way, the traffic conditions on all the road segments of a transportation network in the time slot t are modeled as a traffic network matrix
S
t
…we predict the traffic network matrix
S
t
n
+
1
.” Zhang teaches: The construction of a traffic network matrix is illustrated in Fig. 1. The traffic flow on the road
segment
l
5,8
and
l
8,5
are 16 mph and 13 mph where the transportation network in the time slot t are modeled as a traffic network matrix and we predict the traffic network matrix which maps to: and determine an output of the first generative model in response to the input for the first generative model, wherein the output of the first generative model characterizes the velocity),
wherein the first generative model includes a first component trained to map the input for the first generative model determined depending on the route information [and the probabilistic variable] to an intermediate output which is an acceleration(Zhang, pgs. 7-9, see also fig. 11, “In order to evaluate the effectiveness of TrafficGAN, we first define the traffic conditions of the road segments based on the average speed of traffic flow. Average speeds of 0-9, 10-14, 15-19, 20-24, and over 25 mph correspond to the traffic conditions of heavy congestion, medium-heavy congestion, medium congestion, light congestion and free flow for the road segments, respectively[wherein the first generative model includes a first component trained to map the input for the first generative model determined depending on the route information]…[i]n order to more clearly display the performance of Traffic-GAN, we compare the prediction results of several preferable models on a same segment at different times of a day. Fig 11 presents the predicted average speeds for TrafficGAN[to an intermediate output which is an acceleration]…as well as the real data. One can see that the prediction of TrafficGAN (red curve) better fits with the curve of the real data (blue curve) and more accurately reflects the variation trend of the average speed.”).20
[and determine a characteristic of the velocity over time depending on]a plurality of inputs for the first generative model and [a plurality of inputs for the second model, wherein a series of values for the velocity is determined as the characteristic of the velocity over time](Zhang, pgs. 7-9, see also fig. 11, “The dataset used for evaluation are collected from Chicago Transit Authority (CTA) buses on Chicago’s arterial streets (non-freeway streets) in real-time by continuously monitoring and analyzing their GPS traces…[i]n order to evaluate the effectiveness of TrafficGAN, we first define the traffic conditions of the road segments based on the average speed of traffic flow[a plurality of inputs for the first generative model and].”).21
While Zhang does teach a first generative model, Zhang does not teach: a probabilistic variable including noise; and the probabilistic variable; a hardware processor that implements.
However, Goodfellow teaches
[provide an input for a first generative model depending on a route information], a probabilistic variable including noise, [and depending on an output of the second physical
model] (Goodfellow, pg. 2, “To learn the generator’s distribution
p
g
over data
x
, we define a prior on input noise variables
p
z
(
z
)
then represent a mapping to data space as G(z;
θ
g
) [where G represent the generator and z represents the sampled noise from the noise distribution
p
z
(
z
)
]…”);22
[wherein the first generative model includes a first component trained to map the input for the first generative model determined depending on the route information] and the probabilistic variable [to an intermediate output which is an acceleration]( Goodfellow, pg. 2, “To learn the generator’s distribution
p
g
over data
x
, we define a prior on input noise variables
p
z
(
z
)
then represent a mapping to data space as G(z;
θ
g
) [where G represent the generator and z represents the sampled noise from the noise distribution
p
z
(
z
)
]…”).23
a hardware processor that implements(Goodfellow, pg. 5, “We trained adversarial nets [on]…a range of datasets including MNIST… [and] the Toronto Face Database….”).24
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang with the teachings of Goodfellow the motivation to do so would be to add random noise to the generator to avoid using approximate inference or Markov chains during training(Goodfellow, pg. 2, “[W]e explore the special case when the generative model generates samples by passing random noise through a multilayer perceptron…[w]e refer to this special case as adversarial nets. In this case, we can train both models using only the highly successful backpropagation and dropout algorithms…and sample from the generative model using only forward propagation…[with] [n]o approximate inference or Markov chains…[being] necessary.”).
While Zhang in view of Goodfellow do teach a first generative model, and the probabilistic variable including noise, Zhang in view of Goodfellow do not teach: a second physical model; and depending on an output of the second physical model.
However, Dietrich teaches:
a second physical model(Dietrich, 292, see also fig. 11.8 (as reproduced herein), “The vehicle model described below supplements the model described in the previous section with wheel suspension kinematics.”
PNG
media_image1.png
277
600
media_image1.png
Greyscale
);25
[providing an input for a first generative model depending on a route information, a probabilistic variable including noise,] and depending on an output of the second physical model(Dietrich, 292, see also fig. 11.8 (as reproduced herein), “The vehicle model described below supplements the model described in the previous section with wheel suspension kinematics.”
PNG
media_image1.png
277
600
media_image1.png
Greyscale
);26, 27
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow with the teachings of Dietrich the motivation to do so would be to create a realistic model to better incorporate the dynamics of a real vehicle for different driving patterns(Dietrich, pgs. 8, “The design and testing of such systems with their enormous variety of functions is a high challenge. Requirements for design methods and testing programs and the resulting modeling and simulation technology... driving maneuvers can be simulated as often and reproducibly as desired under defined boundary conditions... critical driving maneuvers can be replaced by safe simulation.”).28
While Zhang in view of Goodfellow and Dietrich do teach a first generative model, a second physical model, and the probabilistic variable including noise, Zhang in view of Goodfellow and Dietrich do not teach: includes a second component trained to map the intermediate output to the velocity of the vehicle depending on an output of a second
model; wherein the output of the second model characterizes a physical constraint for the
intermediate output, wherein the physical constraint is based on the route information;
provide an input for the second model depending on at least one vehicle state or
the route information; determine the output of the second model in response to the input for the second model; and determine a characteristic of the velocity over time depending on a plurality of inputs for the second model, wherein a series of values for the velocity is determined as the characteristic of the velocity over time
However, Shih teaches:
[wherein the first generative model] includes a second component trained to map the
intermediate output to the velocity of the vehicle depending on an output of a second
model(Shih, pgs., 371-374, see also figs. 6 and 9, “Figure 9 illustrates the flow of the proposed EDA network, which consists of four major steps… [t]he model takes a sequence of velocity. Each velocity, V , is fed into a uniform quantization with 256 classes… [a]fter the quantization step, one-hot encoding is conducted so as to take a sequence of velocity and generate a prediction for 5 seconds… there are 50 points for 5 seconds for each input sequence, shown in Eqn. 2, and output sequence, shown in Eqn. 3… [a]t the end of the flow, we need to do inverse quantization for calculation of MSE.”);29
wherein the output of the second model characterizes a physical constraint for the
intermediate output, wherein the physical constraint is based on the route information(Shih, pgs., 371-374, see also figs. 6 and 9, “The proposed EDA networks, shown in Figure 6, starts with data classifier to determine the driving scenario and continues to feed the aligned input sequence into prediction networks… [t]he data classifier annotates the input sequences for different driving scenarios so that the subsequent steps can use different models to predict the vehicle velocity… [for] [t]he second classifier annotates the speed limit at the vehicle location, which are available from navigation map services. In this work, the algorithm
query openstreetmap to obtain the road types and speed limit based on the vehicle location.”);
provide an input for the second model depending on at least one vehicle state or
the route information(Shih, pgs., 371-374, see also figs. 6 and 9, “The proposed EDA networks, shown in Figure 6, starts with data classifier to determine the driving scenario and continues to feed the aligned input sequence into prediction networks…[for] [t]he second classifier annotates the speed limit at the vehicle location, which are available from navigation map services. In this work, the algorithm query openstreetmap to obtain the road types and speed limit based on the vehicle location. For each road segment of the trajectory, the algorithm calculates the average speed of the vehicle and annotates it into five classes. When compared with the speed limit, the average speed class will serve as an indication of driver intention and traffic condition.”);
determine the output of the second model in response to the input for the second
model; and determine a characteristic of the velocity over time depending on [a plurality of inputs for the first generative model and] a plurality of inputs for the second model, wherein a series of values for the velocity is determined as the characteristic of the velocity over time(Shih, pgs., 373-375, see also Table III, “Two datasets are used for the sake of generosity… [t]he first dataset is a private dataset collected from nine bus routes… [t]he collected data include: value from gyroscope in x, y, z directions at 10Hz, acceleration in x,y,z direction at 10Hz, latitude and longitude coordinates at 0.5Hz,GPS speed at 0.5Hz, wheel speed at 2Hz… [t]he time span of each route ranges from 2min to 20min… [h]ence, by detecting the change of lane ID, we could annotate the lane change. Last, the data in NGSIM dataset are all collected on highway… [i]n the experiment, the three maneuver classes, including change lane, go straight, and turn, are compared individually…[t]able III shows the predication errors in private dataset while EDA network model is used. The results show that the
PNG
media_image2.png
171
530
media_image2.png
Greyscale
model does have different performance for different driving behaviors. When the vehicle always drive straight, the model predicts best; when the vehicle changes lane or makes turn, the model leads to higher predication errors although the difference ranges from 0.09 meter per second to 0.11 meter per second”).30
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow and Dietrich with the teachings of Shih the motivation to do so would be to devise a prediction system to avoid collusions among vehicles(Shih, pg., 369, “Free space defines the geometry region for a vehicle to safely move in a certain time period and is essential for every vehicle, controlled by both human and computer software, to prevent collision. Over-conservative estimation for free space will not only reduce the utilization of the shared space but also lead to traffic accident. Motion prediction of nearby vehicles are essential to estimate free space and is the goal of this work.”).
Zhang in view of Goodfellow, Dietrich, and Shih do not teach: a third model.
However, Yufang teaches:
a third model(Yufang, pgs. 1285-1286, see also figs. 7 and 8, “The long-term vehicle speed prediction algorithm designed in this paper is characteri[z]ed by the ability to select different suitable algorithms to predict the speed of the vehicle according to the type of road segment, involving BP algorithm prediction on Known/Unknown city roads, LSTM algorithm prediction on Known/Unknown suburb roads and freeway roads. Among them, the predicted vehicle speed can basically keep up with the change law of the real vehicle speed, as is shown in Figs. 7 and 8….”Yufang teaches: The long-term vehicle speed prediction algorithm designed in this paper is characteri[z]ed by the ability to select different suitable algorithms to predict the speed of the vehicle according to the type of road segment, involving BP algorithm prediction on Known/Unknown city roads, LSTM algorithm prediction on Known/Unknown suburb roads and freeway roads which maps to: a third model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, Dietrich, and Shih with the teachings of Yufang the motivation to do so would be to predict a vehicle’s speed using a hybrid neural network that is able to learn the underlying behavior of drivers for better time prediction and energy consumption(Yufang, pg. 1281, “[B]y comparing the efficiency and rationality of the several algorithms, a novel data-driven back propagation-long short-term memory (BP-LSTM) algorithm for on-road individual long-term average vehicle speed prediction along a driving route is proposed…[t]he novelty of the proposed method lies in the consideration of unobservable driving states in the prediction model hidden in the collected individual driving data, such as the driver's behavio[]rs… the long-term vehicle speed prediction result in this paper can be used well for travel time prediction and energy consumption prediction scenarios, which can reali[z]e the intelligent navigation and energy saving for new energy vehicles.”).
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Zhang, Yuxuan, et al. "Trafficgan: Network-scale deep traffic prediction with generative adversarial nets." IEEE Transactions on Intelligent Transportation Systems 22.1 (2019) (“Zhang”) in view of Goodfellow, Ian, et al. "Generative adversarial nets." Advances in neural information processing systems 27 (2014)(“Goodfellow”) and in view of Dieter et al., “Modellbildung und Simulation der Dynamik von Kraftfahrzeugen,” springer, Berlin/Heidelberg, 2010(“Dietrich”) and in view of Shih, Chi-Sheng, et al. "Vehicle speed prediction with RNN and attention model under multiple scenarios." 2019 IEEE Intelligent Transportation Systems Conference (ITSC). IEEE, 2019(“Shih”) and in view of Yufang et al., "Investigating long‐term vehicle speed prediction based on BP‐LSTM algorithms." IET Intelligent Transport Systems 13.8 (2019) (“Yufang”) and further in view of CN106650287 (A)(“Kang”).
Regarding claim 8, Zhang in view of Goodfellow, Dietrich, Shih and Yufang teaches the method according to claim 1, but does not teach further comprising: estimating an exhaust characteristic for the vehicle depending on the characteristic of velocity over time and/or the score.
However, Kang teaches: estimating an exhaust characteristic for the vehicle depending on the characteristic of velocity over time and/or the score(Kang, pg. 1, abstract, “The method includes the following steps that monitor vehicle exhaust…of motor vehicles on an actual road…[using] relevant data include[ing] the type, speed and acceleration of the motor vehicles…”).31
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Zhang in view of Goodfellow, Dietrich, Shih and Yufang with the teachings of Kang the motivation to do so would be to get real-time estimations of emission data to track the amount of chemicals being released when driving(Kang, pg. 1, abstract, “[B]ased on the emission factor database of the motor vehicle exhaust CO, HC, and NO and the relevant data collected by the motor vehicle exhaust remote-sensing monitoring equipment, MLP neural network models for CO, HC and NO are established respectively. Thus, real-time online estimation of motor vehicle exhaust emission factors can be achieved.”).
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.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ADAM C STANDKE whose telephone number is (571)270-1806. The examiner can normally be reached Gen. M-F 9-9PM EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael J Huntley can be reached at (303) 297-4307. 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.
/Adam C Standke/
Primary Examiner
Art Unit 2129
1 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Zhang
2 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Zhang
3 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Zhang
4 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Goodfellow
5 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Goodfellow
6 Untranslated-Version: Das nachfolgende beschriebene Fahrzeugmodell ergänzt das im vorigen Abschnitt beschriebene Modell um eine Radaufhängungskinematik.
7 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Dietrich
8 Untranslated-Version: Die Auslegung and Erprobung derartiger System emit ihrer enormer Funktionsvielfalt stellt hohe. Anforderungen an Auslegungsmethoden und Erprobunsprogramme and daraus resultierend die Modellierungs and Simulationstechnik…[f]ahrman
o
¨
ver lasses sich unter definierten Randbedingungen beliebig oft und reproduzierbar simulieren…[k]ritische Fahrman
o
¨
ver konnen durch gefahrlose Simulation ersetzt werden
9 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Shih
10 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Shih
11 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
12 Untranslated-Version: Die Kraftangriffspunkte der Aufbaufeder und des Aufbaudämpfers sollen im Rahmen einer Parametrierung des Modells noch frei festlegbar sein. Aus diesem Grund werden für die feder and den Dämpfer jeweils unterschiedliche Angriffspunkte der Kraftelemente festgelegt.
13 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
14 Untranslated-Version: Aufbauend auf den Gln. (11.80) bis (11.83) lassen sich die aboluten Geschwindigkeiten…angeben
15 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
16 Because MNIST and the Toronto Face Database is software that requires a computer for execution it is inherent
that a computer contains a processor and a memory within.
17 Untranslated-Version: In Tabelle 11.8 sind exemplarish einige Werte für die beschriebenen Fahrzeugmodelle angegeben
18 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
19 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Zhang
20 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Zhang
21 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Zhang
22 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Goodfellow
23 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Goodfellow
24 Examiner Remarks: Because MNIST and the Toronto Face Database is software that requires a computer for execution it is inherent that a computer contains a processor and a memory within.
25 Untranslated-Version: Das nachfolgende beschriebene Fahrzeugmodell ergänzt das im vorigen Abschnitt beschriebene Modell um eine Radaufhängungskinematik.
26 Untranslated-Version: Das nachfolgende beschriebene Fahrzeugmodell ergänzt das im vorigen Abschnitt beschriebene Modell um eine Radaufhängungskinematik.
27 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Dietrich
28 Untranslated-Version: Die Auslegung and Erprobung derartiger System emit ihrer enormer Funktionsvielfalt stellt hohe. Anforderungen an Auslegungsmethoden und Erprobunsprogramme and daraus resultierend die Modellierungs and Simulationstechnik…[f]ahrman
o
¨
ver lasses sich unter definierten Randbedingungen beliebig oft und reproduzierbar simulieren…[k]ritische Fahrman
o
¨
ver konnen durch gefahrlose Simulation ersetzt werden
29 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Shih
30 Examiner Remarks: Claim limitations that are not in bold and contained in square brackets are claim limitations not taught by the prior art of Shih
31 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.