DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
This action is reply to the Application Number 17/911,244 filed on 08/07/2025
Claims 13 – 20 are currently pending and have been examined. Claims 13 – 17 and 20 have been amended.
This action is made NON-FINAL
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 08/07/2025 has been entered.
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.
Claims 13 – 16 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (CN109334378A), further in view of Lakshmanan et al. (US 20190050729 A1).
Regarding claim 13, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising: (Yang: Page 1, lines 16 – 18: “The invention belongs to vehicle suspension system control fields, especially for the vehicle ISD of the used container technique of application (Inerter-Spring-Damper) suspension system controls.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general name of all force transmission connections between vehicle body and wheel,”,
Supplemental Note: the damper is equivalent to the vehicle suspension system)
a controller which includes: (Yang: Page 1, lines 46 – 48: “Based on the above reasons, the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system”)
a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and (Yang: Page 3, lines 6 – 14: “Further, the learning flow of the learning algorithm of step 4) is: 4.1) determining an initial weight coefficient wi(0); 4.2) calculating an error between an actual output and a desired output at a current time; 4.3) if the error is less than The fixed value ends, otherwise it continues; 4.4) the weight coefficient is updated by the learning rule; 4.5) returns to step 4.2). The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface.”,
Supplemental Note: the PID controller and its components are used to perform the calculation process that is connected to the suspension system)
a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, (Yang: Page 3, lines 44 – 45: “1 is a flow chart of a method for actively controlling a vehicle ISD suspension based on a single neuron PID control, and FIG. 2 is a suspension structure in the method example”,
Supplemental Note: the PID controller is part of the suspension structure that is able to apply the target amount onto the suspension structure)
wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure …, wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).”; Page 2, line 22 – Page 3, line 4: “Step 4), determine the neuron learning rules as supervised Delta learning rules:
PNG
media_image1.png
38
325
media_image1.png
Greyscale
Where rj(k) is the expected output, 𝛗j(k) is the activation value of neuron, 𝛗i(k) is the difference between the expected output rj(k) and the actual output 𝛗j(k), and wij(k) is the weight coefficient between the two neurons. The amount of change, d is the learning speed; Then determine the specific learning algorithm according to the determined learning rules:
PNG
media_image2.png
445
650
media_image2.png
Greyscale
Where e(k) is the acceleration error signal, xi(k)(i=1, 2, 3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the corresponding neuron input xi(k) The weighting system, di(i=1,2,3) is the learning speed of wi(i=1,2,3), K is the neuron gain coefficient, and u(k) is the output of the single neuron PID control. signal; The control flow of the single neuron PID control is: taking the desired vehicle body acceleration r(k) as the input signal, the body acceleration as the feedback signal y(k), and e(k) as the acceleration error signal, ie e(k)=r(k) )-y(k), xi(k)(i=1, 2, 3) are the input signals of the neurons obtained by the conversion of the acceleration error, and wi(k) is the weight of the input xi(k) of the corresponding neuron. The coefficient is adjusted online by the determined learning rule and the learning algorithm, and u(k) is an output signal of the single neuron PID control for controlling the output force of the linear motor.”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation. Furthermore, multiple neurons are used for these calculations, thus interpreted as a multilayer structure by one with knowledge in the art)
an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, and wherein the weight coefficient acquisition portion includes a first map indicating a relationship between the predetermined condition output from a mode switch and a gain scheduling parameter, and a second map indicating a relationship between the weight coefficient and the gain scheduling parameter. (Yang: Page 3, lines 16 – 26: “The schematic diagram of the single neuron PID controller is shown in Figure 4. The expected vehicle acceleration r(k) is used as the input signal, the body acceleration is used as the feedback signal y(k), and e(k) is the acceleration error signal, ie e(k). =r(k)-y(k), xi(k)(i=1,2,3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the input of the corresponding neuron xi(k) The weight coefficient is adjusted online by a certain learning rule to realize the dynamic control effect of the single neuron PID controller, K is the neuron gain coefficient, and u(k) is the output signal of the single neuron PID control, which is a straight line. The output force of the motor, the linear motor output force u(k) will change with the change of the road surface input, and the purpose of adjusting the electrical impedance will be changed. The total impedance of the liquid-electric coupling type inertial container will also change with the change of the electrical impedance. When the total impedance matches the input impedance, the performance of the ISD suspension is greatly improved.”; Page 6, lines 10 – 13: “As shown in FIG. 5, it is a comparison diagram between the active controllable ISD performance of the vehicle and the performance of the conventional passive suspension according to the embodiment of the present invention, wherein (a) is the time domain map of the vehicle body acceleration, and (b) is the dynamic deflection of the suspension. The domain map, (c) is the time domain map of the tire dynamic load”,
Supplemental Note: the acceleration of the vehicle and the neuron gain coefficient (equivalent to the gain scheduling parameter) is used by the formula in conjunction with the weight coefficients to calculate the specific weight coefficient needed. These relate to the first and second map as the predetermined conditions regarding the acceleration and the road surface along with the neuron gain coefficient and the weight coefficients are all used in conjunction to find the specific weight coefficient needed)
In sum, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising a controller which includes: a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure, wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, and wherein the weight coefficient acquisition portion includes a first map indicating a relationship between the predetermined condition output from a mode switch and a gain scheduling parameter, and a second map indicating a relationship between the weight coefficient and the gain scheduling parameter. Yang however does not teach the multilayer structure having four or more layers whereas Lakshmanan does.
Lakshmanan teaches including four or more layers, (Lakshmanan: Abstract: “Methods and apparatus relating to deep learning solutions for safe, legal, and/or efficient autonomous driving are described.”; Paragraph 0011: “FIG. 10A-10B illustrate layers of exemplary deep neural networks.”; Paragraph 0069: “The exemplary neural networks described above can be used to perform deep learning. Deep learning is machine learning using deep neural networks. The deep neural networks used in deep learning are artificial neural networks composed of multiple hidden layers, as opposed to shallow neural networks that include only a single hidden layer. Deeper neural networks are generally more computationally intensive to train. However, the additional hidden layers of the network enable multistep pattern recognition that results in reduced output error relative to shallow machine learning techniques.”; Paragraph 0070: “Deep neural networks used in deep learning typically include a front-end network to perform feature recognition coupled to a back-end network which represents a mathematical model that can perform operations (e.g., object classification, speech recognition, etc.) based on the feature representation provided to the model. Deep learning enables machine learning to be performed without requiring hand crafted feature engineering to be performed for the model. Instead, deep neural networks can learn features based on statistical structure or correlation within the input data. The learned features can be provided to a mathematical model that can map detected features to an output. The mathematical model used by the network is generally specialized for the specific task to be performed, and different models will be used to perform different task.”)
PNG
media_image3.png
489
1136
media_image3.png
Greyscale
Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the invention disclosed by Yang with the teachings of Lakshmanan with a reasonable expectation of success. Lakshmanan teaches the ability of an autonomous vehicle to utilize deep learning for safe and efficient autonomous driving. The deep learning technique utilizes multiple layers which increase multistep pattern recognition resulting in fewer errors. Yang teaches the ability of utilizing a neural network to control the suspension of the vehicle by dynamically evaluating different vehicle parameters using neurons. One with knowledge in the art would find it obvious to try to implement the deep learning multilayer structure of Lakshmanan with the vehicle system of Yang. This combination allows Yang to increase the efficiency and decrease the rate of errors when evaluating different vehicle parameters by the use of the multistep pattern recognition as taught by Lakshmanan. This increases the performance of the PID controllers in determining an electrical impedance to apply based on the vehicle parameters evaluated by the multilayer structure, which in-turn improves the ISD suspension system.
Regarding claim 14, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising: (Yang: Page 1, lines 16 – 18: “The invention belongs to vehicle suspension system control fields, especially for the vehicle ISD of the used container technique of application (Inerter-Spring-Damper) suspension system controls.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general name of all force transmission connections between vehicle body and wheel,”,
Supplemental Note: the damper is equivalent to the vehicle suspension system)
a controller which includes: (Yang: Page 1, lines 46 – 48: “Based on the above reasons, the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system”)
a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and (Yang: Page 3, lines 6 – 14: “Further, the learning flow of the learning algorithm of step 4) is: 4.1) determining an initial weight coefficient wi(0); 4.2) calculating an error between an actual output and a desired output at a current time; 4.3) if the error is less than The fixed value ends, otherwise it continues; 4.4) the weight coefficient is updated by the learning rule; 4.5) returns to step 4.2). The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface.”,
Supplemental Note: the PID controller and its components are used to perform the calculation process that is connected to the suspension system)
a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, (Yang: Page 3, lines 44 – 45: “1 is a flow chart of a method for actively controlling a vehicle ISD suspension based on a single neuron PID control, and FIG. 2 is a suspension structure in the method example”,
Supplemental Note: the PID controller is part of the suspension structure that is able to apply the target amount onto the suspension structure)
wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure…, wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).”; Page 2, line 22 – Page 3, line 4: “Step 4), determine the neuron learning rules as supervised Delta learning rules:
PNG
media_image1.png
38
325
media_image1.png
Greyscale
Where rj(k) is the expected output, 𝛗j(k) is the activation value of neuron, 𝛗i(k) is the difference between the expected output rj(k) and the actual output 𝛗j(k), and wij(k) is the weight coefficient between the two neurons. The amount of change, d is the learning speed;
Then determine the specific learning algorithm according to the determined learning rules:
PNG
media_image2.png
445
650
media_image2.png
Greyscale
Where e(k) is the acceleration error signal, xi(k)(i=1, 2, 3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the corresponding neuron input xi(k) The weighting system, di(i=1,2,3) is the learning speed of wi(i=1,2,3), K is the neuron gain coefficient, and u(k) is the output of the single neuron PID control. signal; The control flow of the single neuron PID control is: taking the desired vehicle body acceleration r(k) as the input signal, the body acceleration as the feedback signal y(k), and e(k) as the acceleration error signal, ie e(k)=r(k) )-y(k), xi(k)(i=1, 2, 3) are the input signals of the neurons obtained by the conversion of the acceleration error, and wi(k) is the weight of the input xi(k) of the corresponding neuron. The coefficient is adjusted online by the determined learning rule and the learning algorithm, and u(k) is an output signal of the single neuron PID control for controlling the output force of the linear motor.”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation. Furthermore, multiple neurons are used for these calculations, thus interpreted as a multilayer structure by one with knowledge in the art)
an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, wherein the target amount is a target damping force, and wherein the control instruction value acquisition portion is a damping force map indicating a relationship between the target damping force and an instruction value to be output to the damper. (Yang: Page 3, lines 16 – 26: “The schematic diagram of the single neuron PID controller is shown in Figure 4. The expected vehicle acceleration r(k) is used as the input signal, the body acceleration is used as the feedback signal y(k), and e(k) is the acceleration error signal, ie e(k). =r(k)-y(k), xi(k)(i=1,2,3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the input of the corresponding neuron xi(k) The weight coefficient is adjusted online by a certain learning rule to realize the dynamic control effect of the single neuron PID controller, K is the neuron gain coefficient, and u(k) is the output signal of the single neuron PID control, which is a straight line. The output force of the motor, the linear motor output force u(k) will change with the change of the road surface input, and the purpose of adjusting the electrical impedance will be changed. The total impedance of the liquid-electric coupling type inertial container will also change with the change of the electrical impedance. When the total impedance matches the input impedance, the performance of the ISD suspension is greatly improved.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general term for all force transmission devices between the body and the wheel. Its components usually include springs, damping and guiding mechanisms.” ,
Supplemental Note: the acceleration of the vehicle and the neuron gain coefficient is used by the formula in conjunction with the weight coefficients to calculate the specific weight coefficient needed. The weight coefficient is used for the calculation of the damping force which is the output of the electrical impedance to the suspension system. As stated, the vehicle suspension system uses damping systems to function, thus the electrical impedance are equivalent to the damping force applied onto the suspension)
In sum, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising: a controller which includes: a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure, wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, wherein the target amount is a target damping force, and wherein the control instruction value acquisition portion is a damping force map indicating a relationship between the target damping force and an instruction value to be output to the damper. Yang however does not teach the multilayer structure including four or more layers whereas Lakshmanan does.
Lakshmanan teaches including four or more layers, (Lakshmanan: Abstract: “Methods and apparatus relating to deep learning solutions for safe, legal, and/or efficient autonomous driving are described.”; Paragraph 0011: “FIG. 10A-10B illustrate layers of exemplary deep neural networks.”; Paragraph 0069: “The exemplary neural networks described above can be used to perform deep learning. Deep learning is machine learning using deep neural networks. The deep neural networks used in deep learning are artificial neural networks composed of multiple hidden layers, as opposed to shallow neural networks that include only a single hidden layer. Deeper neural networks are generally more computationally intensive to train. However, the additional hidden layers of the network enable multistep pattern recognition that results in reduced output error relative to shallow machine learning techniques.”; Paragraph 0070: “Deep neural networks used in deep learning typically include a front-end network to perform feature recognition coupled to a back-end network which represents a mathematical model that can perform operations (e.g., object classification, speech recognition, etc.) based on the feature representation provided to the model. Deep learning enables machine learning to be performed without requiring hand crafted feature engineering to be performed for the model. Instead, deep neural networks can learn features based on statistical structure or correlation within the input data. The learned features can be provided to a mathematical model that can map detected features to an output. The mathematical model used by the network is generally specialized for the specific task to be performed, and different models will be used to perform different task.”)
PNG
media_image3.png
489
1136
media_image3.png
Greyscale
Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the invention disclosed by Yang with the teachings of Lakshmanan with a reasonable expectation of success. Please refer to the rejection of claim 13 as both state the same language and therefore rejected under the same pretenses.
Regarding claim 15, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising: (Yang: Page 1, lines 16 – 18: “The invention belongs to vehicle suspension system control fields, especially for the vehicle ISD of the used container technique of application (Inerter-Spring-Damper) suspension system controls.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general name of all force transmission connections between vehicle body and wheel,”,
Supplemental Note: the damper is equivalent to the vehicle suspension system)
a controller which includes: (Yang: Page 1, lines 46 – 48: “Based on the above reasons, the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system”)
a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and (Yang: Page 3, lines 6 – 14: “Further, the learning flow of the learning algorithm of step 4) is: 4.1) determining an initial weight coefficient wi(0); 4.2) calculating an error between an actual output and a desired output at a current time; 4.3) if the error is less than The fixed value ends, otherwise it continues; 4.4) the weight coefficient is updated by the learning rule; 4.5) returns to step 4.2). The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface.”,
Supplemental Note: the PID controller and its components are used to perform the calculation process that is connected to the suspension system)
a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, (Yang: Page 3, lines 44 – 45: “1 is a flow chart of a method for actively controlling a vehicle ISD suspension based on a single neuron PID control, and FIG. 2 is a suspension structure in the method example”,
Supplemental Note: the PID controller is part of the suspension structure that is able to apply the target amount onto the suspension structure)
wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure…, wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).”; Page 2, line 22 – Page 3, line 4: “Step 4), determine the neuron learning rules as supervised Delta learning rules:
PNG
media_image1.png
38
325
media_image1.png
Greyscale
Where rj(k) is the expected output, 𝛗j(k) is the activation value of neuron, 𝛗i(k) is the difference between the expected output rj(k) and the actual output 𝛗j(k), and wij(k) is the weight coefficient between the two neurons. The amount of change, d is the learning speed;
Then determine the specific learning algorithm according to the determined learning rules:
PNG
media_image2.png
445
650
media_image2.png
Greyscale
Where e(k) is the acceleration error signal, xi(k)(i=1, 2, 3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the corresponding neuron input xi(k) The weighting system, di(i=1,2,3) is the learning speed of wi(i=1,2,3), K is the neuron gain coefficient, and u(k) is the output of the single neuron PID control. signal; The control flow of the single neuron PID control is: taking the desired vehicle body acceleration r(k) as the input signal, the body acceleration as the feedback signal y(k), and e(k) as the acceleration error signal, ie e(k)=r(k) )-y(k), xi(k)(i=1, 2, 3) are the input signals of the neurons obtained by the conversion of the acceleration error, and wi(k) is the weight of the input xi(k) of the corresponding neuron. The coefficient is adjusted online by the determined learning rule and the learning algorithm, and u(k) is an output signal of the single neuron PID control for controlling the output force of the linear motor.”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation. Furthermore, multiple neurons are used for these calculations, thus interpreted as a multilayer structure by one with knowledge in the art)
an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, and wherein the instruction value acquisition portion includes a front wheel instruction value acquisition portion for a front wheel included in the wheel, and a rear wheel instruction value acquisition portion for a rear wheel included in the wheel. (Yang; Page 3, lines 16 – 26: “The schematic diagram of the single neuron PID controller is shown in Figure 4. The expected vehicle acceleration r(k) is used as the input signal, the body acceleration is used as the feedback signal y(k), and e(k) is the acceleration error signal, ie e(k). =r(k)-y(k), xi(k)(i=1,2,3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the input of the corresponding neuron xi(k) The weight coefficient is adjusted online by a certain learning rule to realize the dynamic control effect of the single neuron PID controller, K is the neuron gain coefficient, and u(k) is the output signal of the single neuron PID control, which is a straight line. The output force of the motor, the linear motor output force u(k) will change with the change of the road surface input, and the purpose of adjusting the electrical impedance will be changed. The total impedance of the liquid-electric coupling type inertial container will also change with the change of the electrical impedance. When the total impedance matches the input impedance, the performance of the ISD suspension is greatly improved.”; Page 5, lines 54 – 56: “The vehicle single-wheel suspension model is established in Matlab/Simulink. According to the above control method, the m function of the fitness function and the genetic algorithm is written. The individual assigns values sequentially and outputs the most fitness function value, as shown in Table 2.”; Page 1, lines 46 – 48: “the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system, and determine an optimal solution of control parameters”,
Supplemental Note: the acceleration of the vehicle and the neuron gain coefficient is used by the formula in conjunction with the weight coefficients to calculate the specific weight coefficient needed. The calculation can be done for each single wheel which is interpreted to be able to acquire values of all wheels on the vehicle, being able to be used on multiple components of the vehicle IDS system)
In sum, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising: a controller which includes: a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure, wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, and wherein the instruction value acquisition portion includes a front wheel instruction value acquisition portion for a front wheel included in the wheel, and a rear wheel instruction value acquisition portion for a rear wheel included in the wheel. Yang however does not teach a multilayer structure including four or more layers whereas Lakshmanan does.
Lakshmanan teaches including four or more layers, (Lakshmanan: Abstract: “Methods and apparatus relating to deep learning solutions for safe, legal, and/or efficient autonomous driving are described.”; Paragraph 0011: “FIG. 10A-10B illustrate layers of exemplary deep neural networks.”; Paragraph 0069: “The exemplary neural networks described above can be used to perform deep learning. Deep learning is machine learning using deep neural networks. The deep neural networks used in deep learning are artificial neural networks composed of multiple hidden layers, as opposed to shallow neural networks that include only a single hidden layer. Deeper neural networks are generally more computationally intensive to train. However, the additional hidden layers of the network enable multistep pattern recognition that results in reduced output error relative to shallow machine learning techniques.”; Paragraph 0070: “Deep neural networks used in deep learning typically include a front-end network to perform feature recognition coupled to a back-end network which represents a mathematical model that can perform operations (e.g., object classification, speech recognition, etc.) based on the feature representation provided to the model. Deep learning enables machine learning to be performed without requiring hand crafted feature engineering to be performed for the model. Instead, deep neural networks can learn features based on statistical structure or correlation within the input data. The learned features can be provided to a mathematical model that can map detected features to an output. The mathematical model used by the network is generally specialized for the specific task to be performed, and different models will be used to perform different task.”)
PNG
media_image3.png
489
1136
media_image3.png
Greyscale
Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the invention disclosed by Yang with the teachings of Lakshmanan with a reasonable expectation of success. Please refer to the rejection of claim 13 as both state the same language and therefore rejected under the same pretenses.
Regarding claim 16, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising: (Yang: Page 1, lines 16 – 18: “The invention belongs to vehicle suspension system control fields, especially for the vehicle ISD of the used container technique of application (Inerter-Spring-Damper) suspension system controls.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general name of all force transmission connections between vehicle body and wheel,”,
Supplemental Note: the damper is equivalent to the vehicle suspension system)
a controller which includes: (Yang: Page 1, lines 46 – 48: “Based on the above reasons, the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system”)
a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and (Yang: Page 3, lines 6 – 14: “Further, the learning flow of the learning algorithm of step 4) is: 4.1) determining an initial weight coefficient wi(0); 4.2) calculating an error between an actual output and a desired output at a current time; 4.3) if the error is less than The fixed value ends, otherwise it continues; 4.4) the weight coefficient is updated by the learning rule; 4.5) returns to step 4.2). The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface.”,
Supplemental Note: the PID controller and its components are used to perform the calculation process that is connected to the suspension system)
a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, (Yang: Page 3, lines 44 – 45: “1 is a flow chart of a method for actively controlling a vehicle ISD suspension based on a single neuron PID control, and FIG. 2 is a suspension structure in the method example”,
Supplemental Note: the PID controller is part of the suspension structure that is able to apply the target amount onto the suspension structure)
wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure…, and wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).”; Page 2, line 22 – Page 3, line 4: “Step 4), determine the neuron learning rules as supervised Delta learning rules:
PNG
media_image1.png
38
325
media_image1.png
Greyscale
Where rj(k) is the expected output, 𝛗j(k) is the activation value of neuron, 𝛗i(k) is the difference between the expected output rj(k) and the actual output 𝛗j(k), and wij(k) is the weight coefficient between the two neurons. The amount of change, d is the learning speed;
Then determine the specific learning algorithm according to the determined learning rules:
PNG
media_image2.png
445
650
media_image2.png
Greyscale
Where e(k) is the acceleration error signal, xi(k)(i=1, 2, 3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the corresponding neuron input xi(k) The weighting system, di(i=1,2,3) is the learning speed of wi(i=1,2,3), K is the neuron gain coefficient, and u(k) is the output of the single neuron PID control. signal; The control flow of the single neuron PID control is: taking the desired vehicle body acceleration r(k) as the input signal, the body acceleration as the feedback signal y(k), and e(k) as the acceleration error signal, ie e(k)=r(k) )-y(k), xi(k)(i=1, 2, 3) are the input signals of the neurons obtained by the conversion of the acceleration error, and wi(k) is the weight of the input xi(k) of the corresponding neuron. The coefficient is adjusted online by the determined learning rule and the learning algorithm, and u(k) is an output signal of the single neuron PID control for controlling the output force of the linear motor.”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation. Furthermore, multiple neurons are used for these calculations, thus interpreted as a multilayer structure by one with knowledge in the art)
an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, further comprising: a feedback target amount acquisition portion configured to acquire a target amount for performing feedback control based on the input vehicle state amount; and a mediation portion configured to acquire the target amount to be output to the control instruction value acquisition portion based on the target amount acquired by the instruction value acquisition portion and the target amount acquired by the feedback target amount acquisition portion. (Yang: Page 3, lines 16 – 26: “The schematic diagram of the single neuron PID controller is shown in Figure 4. The expected vehicle acceleration r(k) is used as the input signal, the body acceleration is used as the feedback signal y(k), and e(k) is the acceleration error signal, ie e(k). =r(k)-y(k), xi(k)(i=1,2,3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the input of the corresponding neuron xi(k) The weight coefficient is adjusted online by a certain learning rule to realize the dynamic control effect of the single neuron PID controller, K is the neuron gain coefficient, and u(k) is the output signal of the single neuron PID control, which is a straight line. The output force of the motor, the linear motor output force u(k) will change with the change of the road surface input, and the purpose of adjusting the electrical impedance will be changed. The total impedance of the liquid-electric coupling type inertial container will also change with the change of the electrical impedance. When the total impedance matches the input impedance, the performance of the ISD suspension is greatly improved.”,
Supplemental Note: the acceleration of the vehicle and the neuron gain coefficient is used by the formula in conjunction with the weight coefficients to calculate the specific weight coefficient needed. This calculation, as stated above is given to the PID controller which is equivalent to both the feedback target amount acquisition portion and the mediation portion as it performs the same claims tasks of acquiring the calculated electrical impedance based on the inputs of the current vehicle state and then to output that force)
In sum, Yang teaches a vehicle control apparatus employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control apparatus comprising; a controller which includes: a calculation processing portion configured to make a predetermined calculation based on an input vehicle state amount and output a target amount; and a control instruction value acquisition portion configured to acquire a control instruction value for controlling the damper based on the target amount, wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure, and wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning, further comprising: a feedback target amount acquisition portion configured to acquire a target amount for performing feedback control based on the input vehicle state amount; and a mediation portion configured to acquire the target amount to be output to the control instruction value acquisition portion based on the target amount acquired by the instruction value acquisition portion and the target amount acquired by the feedback target amount acquisition portion. Yang however does not teach a multilayer structure including four or more layers whereas Lakshmanan does.
Lakshmanan teaches including four or more layers, (Lakshmanan: Abstract: “Methods and apparatus relating to deep learning solutions for safe, legal, and/or efficient autonomous driving are described.”; Paragraph 0011: “FIG. 10A-10B illustrate layers of exemplary deep neural networks.”; Paragraph 0069: “The exemplary neural networks described above can be used to perform deep learning. Deep learning is machine learning using deep neural networks. The deep neural networks used in deep learning are artificial neural networks composed of multiple hidden layers, as opposed to shallow neural networks that include only a single hidden layer. Deeper neural networks are generally more computationally intensive to train. However, the additional hidden layers of the network enable multistep pattern recognition that results in reduced output error relative to shallow machine learning techniques.”; Paragraph 0070: “Deep neural networks used in deep learning typically include a front-end network to perform feature recognition coupled to a back-end network which represents a mathematical model that can perform operations (e.g., object classification, speech recognition, etc.) based on the feature representation provided to the model. Deep learning enables machine learning to be performed without requiring hand crafted feature engineering to be performed for the model. Instead, deep neural networks can learn features based on statistical structure or correlation within the input data. The learned features can be provided to a mathematical model that can map detected features to an output. The mathematical model used by the network is generally specialized for the specific task to be performed, and different models will be used to perform different task.”)
PNG
media_image3.png
489
1136
media_image3.png
Greyscale
Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the invention disclosed by Yang with the teachings of Lakshmanan with a reasonable expectation of success. Please refer to the rejection of claim 13 as both state the same language and therefore rejected under the same pretenses.
Regarding claim 20, Yang teaches a vehicle control method employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control method comprising: (Yang: Page 1, lines 16 – 18: “The invention belongs to vehicle suspension system control fields, especially for the vehicle ISD of the used container technique of application (Inerter-Spring-Damper) suspension system controls.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general name of all force transmission connections between vehicle body and wheel,”,
Supplemental Note: the damper is equivalent to the vehicle suspension system)
a controller which includes: (Yang: Page 1, lines 46 – 48: “Based on the above reasons, the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system”)
a calculation processing step of making a predetermined calculation based on an input vehicle state amount and outputting a target amount; and (Yang: Page 3, lines 6 – 14: “Further, the learning flow of the learning algorithm of step 4) is: 4.1) determining an initial weight coefficient wi(0); 4.2) calculating an error between an actual output and a desired output at a current time; 4.3) if the error is less than The fixed value ends, otherwise it continues; 4.4) the weight coefficient is updated by the learning rule; 4.5) returns to step 4.2). The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface.”,
Supplemental Note: the PID controller and its components are used to perform the calculation process that is connected to the suspension system)
a control instruction value acquisition step of acquiring a control instruction value for controlling the damper based on the target amount, (Yang: Page 3, lines 44 – 45: “1 is a flow chart of a method for actively controlling a vehicle ISD suspension based on a single neuron PID control, and FIG. 2 is a suspension structure in the method example”,
Supplemental Note: the PID controller is part of the suspension structure that is able to apply the target amount onto the suspension structure)
wherein the calculation processing step includes making the calculation using a learning result that is acquired by causing a calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure…, and wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation)
an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning. (Yang: Page 3, lines 16 – 26: “The schematic diagram of the single neuron PID controller is shown in Figure 4. The expected vehicle acceleration r(k) is used as the input signal, the body acceleration is used as the feedback signal y(k), and e(k) is the acceleration error signal, ie e(k). =r(k)-y(k), xi(k)(i=1,2,3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the input of the corresponding neuron xi(k) The weight coefficient is adjusted online by a certain learning rule to realize the dynamic control effect of the single neuron PID controller, K is the neuron gain coefficient, and u(k) is the output signal of the single neuron PID control, which is a straight line. The output force of the motor, the linear motor output force u(k) will change with the change of the road surface input, and the purpose of adjusting the electrical impedance will be changed. The total impedance of the liquid-electric coupling type inertial container will also change with the change of the electrical impedance. When the total impedance matches the input impedance, the performance of the ISD suspension is greatly improved.”,
Supplemental Note: the acceleration of the vehicle and the neuron gain coefficient is used by the formula in conjunction with the weight coefficients to calculate the specific weight coefficient needed)
In sum, Yang teaches a vehicle control method employed for a vehicle including a damper configured to adjust a force between a vehicle body of the vehicle and a wheel of the vehicle, the vehicle control method comprising: a controller which includes: a calculation processing step of making a predetermined calculation based on an input vehicle state amount and outputting a target amount; and a control instruction value acquisition step of acquiring a control instruction value for controlling the damper based on the target amount, wherein the calculation processing step includes making the calculation using a learning result that is acquired by causing a calculation processing portion to learn pairs of a plurality of target amounts acquired using a predetermined evaluation method prepared in advance with respect to a plurality of different vehicle state amounts as pairs of pieces of input and output data, wherein the learning result is a weight coefficient acquired by deep learning, which uses a neural network for the learning, the neural network having a multilayer structure…, and wherein the calculation processing portion includes a weight coefficient acquisition portion configured in such a manner that a plurality of different weight coefficients acquired by applying the deep learning using a plurality of different weights is set thereto, the weight coefficient acquisition portion being configured to acquire a specific weight coefficient among the plurality of different weight coefficients according to an input predetermined condition, and an instruction value acquisition portion configured in such a manner that the specific weight coefficient is set thereto, the instruction value acquisition portion being configured to make the calculation that uses the neural network for learning. Yang however does not teach a multilayer structure including four or more layers whereas Lakshmanan does.
Lakshmanan teaches including four or more layers, (Lakshmanan: Abstract: “Methods and apparatus relating to deep learning solutions for safe, legal, and/or efficient autonomous driving are described.”; Paragraph 0011: “FIG. 10A-10B illustrate layers of exemplary deep neural networks.”; Paragraph 0069: “The exemplary neural networks described above can be used to perform deep learning. Deep learning is machine learning using deep neural networks. The deep neural networks used in deep learning are artificial neural networks composed of multiple hidden layers, as opposed to shallow neural networks that include only a single hidden layer. Deeper neural networks are generally more computationally intensive to train. However, the additional hidden layers of the network enable multistep pattern recognition that results in reduced output error relative to shallow machine learning techniques.”; Paragraph 0070: “Deep neural networks used in deep learning typically include a front-end network to perform feature recognition coupled to a back-end network which represents a mathematical model that can perform operations (e.g., object classification, speech recognition, etc.) based on the feature representation provided to the model. Deep learning enables machine learning to be performed without requiring hand crafted feature engineering to be performed for the model. Instead, deep neural networks can learn features based on statistical structure or correlation within the input data. The learned features can be provided to a mathematical model that can map detected features to an output. The mathematical model used by the network is generally specialized for the specific task to be performed, and different models will be used to perform different task.”)
PNG
media_image3.png
489
1136
media_image3.png
Greyscale
Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the invention disclosed by Yang with the teachings of Lakshmanan with a reasonable expectation of success. Please refer to the rejection of claim 13 as both state the same language and therefore rejected under the same pretenses.
Claims 17 – 19 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (CN109334378A) and Lakshmanan et al. (US 20190050729 A1) as applied to claim 13 above, and further in view of Ishioka et al. (US 5409078 A).
Regarding claim 17, Yang, as modified, teaches wherein the predetermined evaluation method includes an evaluation function, and (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).” Page 1, lines 16 – 18: “The invention belongs to vehicle suspension system control fields, especially for the vehicle ISD of the used container technique of application (Inerter-Spring-Damper) suspension system controls.”;
Page 1, lines 46 – 48: “Based on the above reasons, the present invention provides a vehicle ISD suspension active control method based on single neuron PID control, which can be used to perform dynamic analysis on multiple components of a vehicle ISD suspension system”
Page 3, lines 6 – 14: “Further, the learning flow of the learning algorithm of step 4) is: 4.1) determining an initial weight coefficient wi(0); 4.2) calculating an error between an actual output and a desired output at a current time; 4.3) if the error is less than The fixed value ends, otherwise it continues; 4.4) the weight coefficient is updated by the learning rule; 4.5) returns to step 4.2). The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface.”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation)
In sum, Yang teaches wherein the predetermined evaluation method includes an evaluation function. Yang however does not teach wherein the evaluation function includes a configuration divided into a low-frequency component for reducing low-frequency vibration of a vertical acceleration of the vehicle and a high-frequency component for reducing high-frequency signals of the vertical acceleration of the vehicle whereas Ishioka does.
Ishioka teaches wherein the evaluation function includes a configuration divided into a low-frequency component for reducing low-frequency vibration of a vertical acceleration of the vehicle and a high-frequency component for reducing high-frequency signals of the vertical acceleration of the vehicle. (Ishioka: Col. 2, lines 13 – 33: “the vibrations of the power unit which cause the vehicle body to vibrate include (1) high-frequency vibrations, such as booming noise, due to explosion process of the engine and rotation of its crankshaft, and (2) low-frequency vibrations, such as shakes, due to the resonance of the power unit suspension system resulting from an oscillating force applied from the vehicle wheels while the automobile is running on a bumpy road surface. The high-frequency vibrations indicated above can be effectively isolated by an adaptive control using the signal from the vibration sensor attached to the power unit, since the high-frequency vibrations detected at virtually all points on the vehicle body are highly closely related to the vibrations of the power unit. On the other hand, when the low-frequency vibrations such as shakes arise from the resonance of the power unit suspension system, an adaptive control of the oscillating means on the basis of the vibration of the power unit will usually result in reduction in the oscillation of the power unit per se.”: Col. 5, line 50 - “In the instant embodiment, the signal "X.sub.1 " representative of the vibration of the power unit as detected by the first vibration sensor 16 and the signal "X.sub.2 " representative of the displacement of the suspension member in response to movements of the wheels 18 are both used as the first and second reference signals. Accordingly, the present support apparatus can provide a high vibration damping effect with respect to high-frequency vibrations that cause booming noise, for example, based on the input signal "X.sub.1 " from the first sensor 16 and at the same time provide a high vibration isolating effect with respect to low-frequency vibrations that cause shakes, for example, based on the input signal "X.sub.2 " from the second sensor 20. Since the effective damping of the low-frequency vibrations such as shakes can only be achieved by reduction of the oscillation of the power unit 10 per se, it is difficult to control the damping effect as desired, based on only the first reference signal "X.sub.1 " corresponding to the vibration of the power unit 10 detected by the first sensor 16. However, the present support apparatus as described hitherto is able to stably provide an excellent damping effect for the low-frequency vibrations, owing to the use of the second reference signal "X.sub.2 " indicative of the vibration of the suspension member which has a close relationship with the low-frequency vibrations, such as shakes.”,
Supplemental Note: the high-frequency signals are taught to be the vibrations of a power unit due to an explosion process of the engine and rotation of its crankshaft. As a vehicle accelerates, it is known to one with knowledge in the art that the engine will operate at higher rotations per minute (RPM) which results in increased vibrations, thus teaching the claim limitation of the claimed high-frequency signals due to vertical acceleration)
Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the invention disclosed by Yang with the teachings of Ishioka with a reasonable expectation of success. Both Yang and Ishioka teach vehicle systems that are able to dampen or greatly reduce the amount of vibration applied onto the vehicle. Yang already teaches the use of a low cutoff frequency to determine the vertical displacement and in turn determine the vibrations applied onto the vehicle (Yang: Page 2, lines 4 – 20: “
PNG
media_image4.png
216
836
media_image4.png
Greyscale
Where zs, zu, zb and zr are the vertical displacements of the vehicle body, the tire, the upper end of the damping and the road surface, u is the force at both ends of the liquid-electric coupling type inertial container, f is the power of the linear motor, and c is the damping coefficient, k For the stiffness coefficient of the suspension, b is the inertia coefficient, ms is the sprung mass, mu is the unsprung mass, and kt is the tire stiffness; Step 2), determine the vertical displacement zr of the road surface, the formula is:
PNG
media_image5.png
86
785
media_image5.png
Greyscale
Where n0 is the reference spatial frequency, w(t) is the white noise with a mean of 0, f0 is the lower cutoff frequency; Gq(n0) is the road roughness coefficient; v is the vehicle speed), Ishioka furthers this ability by also acquiring high and low frequencies for these vibration calculations. The high frequency refers to the vibrations put onto the vehicle by the engine of the vehicle while the low frequencies are from the vehicle driving over the road surface. One with knowledge in the art would find it obvious to try to implement both frequencies as taught by Ishioka with the vehicle system of Yang. This combination allows for also detecting vibrations in the engine, of which Yang does not account for, thus adding this high frequency factor into the equation as well increases the efficiency of the vehicle as it can reduce vibrations that are not only part of the low cutoff range.
Regarding claim 18, Yang, as modified, teaches wherein the target amount is a target damping force, and wherein the control instruction value acquisition portion is a damping force map indicating a relationship between the target damping force and an instruction value to be output to the damper. (Yang: Page 3, lines 16 – 26: “The schematic diagram of the single neuron PID controller is shown in Figure 4. The expected vehicle acceleration r(k) is used as the input signal, the body acceleration is used as the feedback signal y(k), and e(k) is the acceleration error signal, ie e(k). =r(k)-y(k), xi(k)(i=1,2,3) is the input signal of the neuron obtained by the conversion of the acceleration error, and wi(k) is the input of the corresponding neuron xi(k) The weight coefficient is adjusted online by a certain learning rule to realize the dynamic control effect of the single neuron PID controller, K is the neuron gain coefficient, and u(k) is the output signal of the single neuron PID control, which is a straight line. The output force of the motor, the linear motor output force u(k) will change with the change of the road surface input, and the purpose of adjusting the electrical impedance will be changed. The total impedance of the liquid-electric coupling type inertial container will also change with the change of the electrical impedance. When the total impedance matches the input impedance, the performance of the ISD suspension is greatly improved.”; Page 1, lines 22 – 23: “Vehicle suspension refers to the general term for all force transmission devices between the body and the wheel. Its components usually include springs, damping and guiding mechanisms.” ,
Supplemental Note: the acceleration of the vehicle and the neuron gain coefficient is used by the formula in conjunction with the weight coefficients to calculate the specific weight coefficient needed. The weight coefficient is used for the calculation of the damping force which is the output of the electrical impedance to the suspension system. As stated the vehicle suspension system uses damping systems to function, thus the electrical impedance are equivalent to the damping force applied onto the suspension)
Regarding claim 19, Yang, as modified, teaches wherein the calculation processing portion is configured to further make the calculation while additionally taking input road surface information into consideration, and (Yang: Page 3, lines 11 – 14: “The invention has the beneficial effects that the single neuron PID controller of the suspension system is designed based on the single neuron theory, and the multi-objective genetic algorithm is used to optimize the control parameters, and the suspension control parameters can be actively adjusted according to different inputs of the random road surface”,
Supplemental Note: the PID controller able to control the suspension uses road surface inputs for its calculations)
wherein the calculation processing portion makes the calculation using a learning result that is acquired by causing the calculation processing portion to learn pairs of a plurality of target amounts acquired using the predetermined evaluation method prepared in advance with respect to the plurality of different vehicle state amounts and a plurality of different pieces of road surface information as pairs of pieces of input and output data. (Yang: Page 3, lines 51 – 54: “Referring to FIG. 1, the vehicle ISD suspension active control method based on single neuron PID control of the present invention comprises: step 1): establishing a suspension system model; step 2): determining road surface input and suspension basic parameters; step 3): Determine the performance evaluation indicator; Step 4): Determine the single neuron PID controller.”; Page 5, lines 11 – 14: “Wherein, the learning flow of the learning algorithm is: 4.1) determining the initial weight coefficient wi(0); 4.2) calculating the error between the actual output and the expected output at the current time; 4.3) if the error is less than the given value, then ending, otherwise Continue; 4.4) update the weight coefficient by learning the rule; 4.5) return to step 4.2).”,
Supplemental Note: the weight coefficient is acquired by a learning algorithm which is equivalent to deep learning. Pairs of the actual and expected output are used to find the weight coefficient. The steps involve in acquiring the current vehicle state and the road conditions data to be used for evaluation)
Response to Arguments
Applicant’s arguments, see section II. Interpretation Under 35 U.S.C. 102(a)(2) and III. Rejections Under 35 U.S.C. 103 of the REMARKS, filed 08/07/2025, with respect to the amendments made to claims 13 – 16 and 20 not being taught by the prior art of Yang or Yang in view of Ishiokia have been fully considered and are persuasive. Examiner agrees that neither Yang or Ishiokia teach the amended claim limitations of a multilayer structure including four or more layers therefore, the rejection has been withdrawn. However, upon further consideration, a new ground of rejection is made in view of Lakshmanan et al. (US 20190050729 A1) which does teach a multilayer structure with four or more layers as amended within claims 13 – 16 and 20.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHIVAM SHARMA whose telephone number is (703)756-1726. The examiner can normally be reached Monday-Friday 8:00-5:00.
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, Erin Bishop can be reached at 571-270-3713. 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.
/SHIVAM SHARMA/ Examiner, Art Unit 3665
/Erin D Bishop/ Supervisory Patent Examiner, Art Unit 3665
Prosecution Insights