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 Office Action is in response to the application filed on 10/14/2025. Claims 2-4, 8, 12-14, and 18 have been cancelled. Claims 1 5-7, 9-11, 15-17, and 19-20 have been amended. Claims 1 5-7, 9-11, 15-17, and 19-20 are presently pending and are presented for examination.
Claim Objections
Claim 15 is objected to because of the following informalities. Appropriate correction is required.
Claim 15 recites are formed of a unit vector…. This is unclear. Examiner will interpret are formed of a unit vector as “comprise unit vectors”.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 6 and 16, the claims recite, in part, a vector space. The specification and drawings show that the vector space can be a subspace of two-dimensional space, see Figs. 6-7 and the specification at page 19 paragraph 2-page 20 paragraph 3. The only subspaces of the set of two-dimensional real numbers are the set containing only the zero vector, infinitely extending lines that pass through the origin, and the entire two-dimensional plane, which is also infinite. The zero vector space or a line through the origin cannot represent any typical driving route, such as those seen in Figs. 6-7. The vector space of the entire two-dimensional plane automatically includes all possible routes, including those specified in dependent claims 6 and 16. This means that dependent claims 6 and 16 do not specify any further limitations to claims 5 and 15, respectively. Appropriate clarification is required. Claims 7 and 17 are rejected for the same reason, as they depend on claims 6 or 16 and do not cure the deficiencies noted above. For purposes of further examination on the merits, Examiner will interpret a vector space configuring the reference driving route as “a set of vectors representing the reference driving route”.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim(s) 1 5,-7, 9-11, 15-17, and 19-20 is/are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. Independent claim 1 is directed toward a machine, and independent claim 11 is directed to a method. Therefore, each of the independent claim(s) 1 and 11 along with the corresponding dependent claims 2-10 and 12-20 are directed to a statutory category of invention under Step 1.
Under Step 2A, Prong 1, the claims are analyzed to determine whether one or more of the claims recites subject matter that falls within one of the following groups of abstract ideas: (1) mental processes, (2) certain methods of organizing human activity, and/or (3) mathematical concepts. In this case, the independent claim(s) 1 and 11 is/are directed to an abstract idea without significantly more. Specifically, the claim(s), under its/their broadest reasonable interpretation(s) cover(s) certain mental processes. The language of independent claim 11 is used for illustration:
A method of detecting a parking position of a vehicle, the method comprising:
generating, a first driving route based on dead reckoning (DR) using speed and a-direction information of the vehicle, wherein the first driving route is configured as a plurality of unit vectors (A person could mentally estimate the route their vehicle had taken while they navigated through an indoor parking lot based on the vehicle’s speed and direction.);
generating an artificial intelligence (AI) model for error correction by performing machine learning;
generating a second driving route by inputting the generated first driving route into the generated AI model, the second driving route being generated by sequentially correcting an error in each of the plurality of unit vectors in chronological order through the AI model (A person could mentally estimate the error in an estimate of their driving route based on other data, for example, their own sense data while driving the route.); and
determining a final parking position of the vehicle in the indoor parking lot based on the generated second driving route (A person could mentally estimate the final parking position based on a corrected representation of their driving route.),
wherein generating the AI model comprises:
generating a plurality of virtual driving route data by reflecting characteristics of actual driving route data in a reference driving route data, wherein at least some of the generated virtual driving route data have a rounded shape in a corner section and a meandering shape in a straight section,
and performing machine learning to train the AI model based on the actual driving route data and the virtual driving route data.
As explained above, independent claim 11 recites at least one abstract idea. The other independent claim(s), claim(s) 1, which is/are similar in scope to claim 11 likewise recite(s) at least one abstract idea under Step 2A, Prong 1.
Under Step 2A, Prong 2, the claims are analyzed to determine whether the claim, as a whole, integrates the abstract idea into a practical application. As noted in the 2019 PEG, it must be determined whether any additional elements in the claim beyond the abstract idea integrate the exception into a practical application in a manner that imposes a meaningful limit on the judicial exception. The courts have indicated that additional elements such as merely using a computer to implement an abstract idea, adding insignificant extra-solution activity, or generally linking use of a judicial exception to a particular technological environment or field of use do not integrate a judicial exception into a “practical application”; see at least MPEP 2106.04(d).
In this case, the mental processes are not integrated into a practical application. The additional limitation(s) of independent claims 1 and 11 amount to implementing the abstract idea on a computer, add insignificant extra-solution activity, and/or generally link use of the judicial exception to a particular technological environment or field of use; see at least MPEP 2106.04(d). More specifically,
generate an artificial intelligence (AI) model… found in independent claim(s) 1 and 11. This limitation amounts to generally linking the use of the abstract idea to a particular technological environment or field of use.
generating a plurality of virtual driving route data by reflecting characteristics of actual driving route data in a reference driving route data, wherein at least some of the generated virtual driving route data have a rounded shape in a corner section … found in independent claim(s) 1 and 11. This limitation amounts to generally linking the use of the abstract idea to a particular technological environment or field of use because it essentially consists of applying standard machine learning techniques at a high level to a specific field of use, i.e. driving routes.
performing machine learning to train the AI model based on the actual driving route data and the virtual driving route data … found in independent claim(s) 1 and 11. This limitation amounts to generally linking the use of the abstract idea to a particular technological environment or field of use because it essentially consists of applying standard machine learning techniques at a high level to a specific field of use, i.e. driving routes.
a processor… found in independent claim 1. This limitation amounts to implementing the abstract idea on a computer.
a memory… found in independent claim 1. This limitation amounts to implementing the abstract idea on a computer.
Therefore, taken alone, the additional elements do not integrate the abstract idea into a practical application. Furthermore, looking at the additional limitation(s) as an ordered combination or as a whole, the limitations add nothing significant that is not already present when looking at the elements taken individually. Because the additional elements do not integrate the abstract idea into a practical application by imposing meaningful limits on practicing the abstract idea, independent claim(s) 1 and 11 is/are directed to an abstract idea.
Under Step 2B, the claims do not include any additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application in Step 2A, Prong Two, the additional element of limiting the use of the idea to one particular environment employs generic computer functions to execute an abstract idea and, therefore, does not add significantly more. Mere instruction to apply an exception using generic computer components or limiting the use of the abstract idea to a particular environment or field of use cannot provide an inventive concept. Additionally, as discussed above, the limitation(s) of claims 1 and 11, as recited above, is/are considered insignificant extra-solution activity.
A conclusion that an additional element is insignificant extra-solution activity in Step 2A must be re-evaluated in Step 2B to determine if the element is more than what is well-understood, routine, and conventional in the field. In this case, the additional limitation of a computer system is well-understood, routine, and conventional activity, because the specification does not provide any indication that the modules described is/are anything more than conventional computer(s).
Because the claims fail to recite anything sufficient to amount to significantly more than the judicial exception, independent claim(s) 1 and 11 is/are patent ineligible under 35 U.S.C. 101.
Dependent claims 5-7, 9-10, 15-17, and 19-20 have been given the full two-part analysis, including analyzing the additional limitations, both individually and in combination. Dependent claims 2-10 and 12-20, when analyzed both individually and in combination, are also patent ineligible under 35 U.S.C. 101 based on the same analysis as above. The additional limitations recited in the dependent claims fail to establish that the dependent claims are not directed to an abstract idea. The additional limitations of the dependent claims, when considered individually and as an ordered combination, do not amount to significantly more than the abstract idea. Accordingly, claims 2-10 and 12-20 are patent ineligible under 35 U.S.C. 101.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 10-11, and 20 are rejected under 35 U.S.C. 103 as being obvious over CN 108871336 A, hereinafter “Yao”, in view of CN 103453913 A, hereinafter “Song”, and NPL document “Understand Data Normalization in Machine Learning”, hereinafter “Zhang”.
Regarding claim 1, Yao discloses An apparatus for detecting a position of a vehicle (See Abstract, the position estimating system uses sensor fusion and angle data to obtain real-time vehicle position), the apparatus comprising:
a processor (See page 2 paragraph 5-page 3 paragraph 1, the system uses received GPS information and vehicle information in a Kalman filter, which is a sequence of mathematical operations. This necessarily takes place on a system comprising a processor with associated memory.); and
a memory storing executable instructions (See page 2 paragraph 5-page 3 paragraph 1, the system uses received GPS information and vehicle information in a Kalman filter, which is a sequence of mathematical operations. This necessarily takes place on a system comprising a processor with associated memory storing the corresponding instructions.) that, when executed by the processor, cause the processor to:
generate a first driving route based on dead reckoning (DR) using speed and a direction information of the vehicle (See page 5 paragraph 7, dead reckoning estimates the position of the vehicle. The sequence of estimated positions of the vehicle estimated by dead reckoning is a first driving route. See page 3 paragraph 8, the plurality of sensors including a speedometer and gyroscope that provides angle, i.e. direction data, are used to perform dead reckoning.);
generate an artificial intelligence (Al) model for error correction by performing machine learning (See page 4 paragraph 5-7, a neural network is trained, i.e. generated, to correct the dead reckoning error.).
generate a second driving route by inputting the generated first driving route into the generated Al model, the second driving route being generated by sequentially correcting an error in each of the plurality of unit vectors in chronological order through the Al model (See page 5 paragraph 7, the dead reckoning system estimates the position of the vehicle. The uncorrected position estimates are the first driving route. See page 5 paragraph 11, the model prediction unit corrects the position estimated by dead reckoning using a neural network, with the neural network taking the position estimated by dead reckoning as input. The sequence of estimated positions of the vehicle estimated by the model prediction unit is a second driving route. See page 7 paragraphs 14-15, the system is used to obtain real-time vehicle estimates. This means that the vectors that make up the first driving route are corrected to produce the second driving route as they are collected, which is sequentially.);
determine a final position of the vehicle based on the generated second driving route (See page 4 paragraph 17-page 5 paragraph 1, an improved estimate of the vehicle’s location is outputted. The last outputted position is a final position of the vehicle based on the generated second driving route, since it is the output of the neural network.); and
wherein in generating the Al model, the processor is configured to:
generate a plurality of virtual driving route data by reflecting characteristics of actual driving route data in a reference driving route data, wherein at least some of the generated virtual driving route data have a rounded shape in a corner section and a meandering shape in a straight section, and perform machine learning to train the AI model based on the actual driving route data and the virtual driving route data (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data and dead reckoning data is used to train the neural network. This is machine learning. The position data is actual driving route data. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training. If the vehicle were driven over roads in a typical manner for any length of time, data from corner and straight road sections would be gathered. Corner sections are inherently driven in a rounded manner in a car, which would be reflected in the gathered position data. Human drivers inherently have an element of meandering, i.e., left and right drift, while driving in straight sections due to imprecision in steering and lateral drift of the automobile. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training. This would cause the neural network’s output, i.e. the virtual route data, to reflect the shape of the gathered data, which in this case would include rounded shapes in corner sections and a meandering shape in straight sections.).
Yao does not explicitly disclose detecting a parking position, wherein the first driving route is configured as a plurality of unit vectors, determining a final parking position of the vehicle in the indoor parking.
Song, in the same field of endeavor and solving a related problem, discloses detecting a parking position (See [0021], the method uses dead reckoning to estimate the position of a vehicle in a parking lot. This is a parking position.) determining a final parking position of the vehicle in the indoor parking (See Abstract, the method estimates the driving route of the vehicle in the indoor parking lot using dead reckoning, an accelerometer, and a gyroscope. See [0021], the method estimates the position of a vehicle in a parking lot. This is a parking position. See [0005], the method updates the estimate using map matching. This is a final parking position.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include estimating the vehicle position in an indoor parking lot and determining the parking position of Song. One of ordinary skill in the art would have been motivated to make this modification in order to accurately record where a car has been parked, as suggested by Song at [0002]. Yao combined with Song does not explicitly disclose wherein the first driving route is configured as a plurality of unit vectors.
Zhang renders obvious wherein the first driving route is configured as a plurality of unit vectors (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Regarding claim 10, Yao combined with Song and Zhang renders obvious the limitations of claim 1. Yao discloses connecting the second driving route to a driving route according to the driving information of the vehicle at the time of arrival to determine the final position of the vehicle (See page 5 paragraph 1, the control unit uses the real-time GPS position when it is available, i.e. a driving route, and the dead reckoning position corrected by the neural network, i.e. the second driving route, when the GPS location is not available, i.e. at the time of arrival. This is sequentially connecting the routes. This continues while the vehicle is in operation and therefore determines the final position of the vehicle.). Song, in the same field of endeavor and solving a related problem, renders obvious at the indoor parking lot (See Abstract, the method estimates the position of a vehicle in an indoor parking lot.); and parking position (See [0021], the method estimates the position of a vehicle in a parking lot. This is a parking position. See [0005], the method updates the estimate using map matching. This is a final parking position.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include estimating the vehicle position in an indoor parking lot and determining the parking position of Song. One of ordinary skill in the art would have been motivated to make this modification in order to accurately record where a car has been parked, as suggested by Song at [0002].
Regarding claim 11, Yao discloses A method detecting a position of a vehicle (See Abstract, the position estimating system uses sensor fusion and angle data to obtain real-time vehicle position), the method comprising:
generating a first driving route based on dead reckoning (DR) using speed and a direction information of the vehicle (See page 5 paragraph 7, dead reckoning estimates the position of the vehicle. The sequence of estimated positions of the vehicle estimated by dead reckoning is a first driving route. See page 3 paragraph 8, the plurality of sensors including a speedometer and gyroscope that provides angle, i.e. direction data, are used to perform dead reckoning.);
generating an artificial intelligence (Al) model for error correction by performing machine learning (See page 4 paragraph 5-7, a neural network is trained, i.e. generated, to correct the dead reckoning error.).
generating a second driving route by inputting the generated first driving route into the generated Al model, the second driving route being generated by sequentially correcting an error in each of the plurality of unit vectors in chronological order through the Al model (See page 5 paragraph 7, the dead reckoning system estimates the position of the vehicle. The uncorrected position estimates are the first driving route. See page 5 paragraph 11, the model prediction unit corrects the position estimated by dead reckoning using a neural network, with the neural network taking the position estimated by dead reckoning as input. The sequence of estimated positions of the vehicle estimated by the model prediction unit is a second driving route. See page 7 paragraphs 14-15, the system is used to obtain real-time vehicle estimates. This means that the vectors that make up the first driving route are corrected to produce the second driving route as they are collected, which is sequentially.);
and determining a final position of the vehicle based on the generated second driving route (See page 4 paragraph 17-page 5 paragraph 1, an improved estimate of the vehicle’s location is outputted. The last outputted position is a final position of the vehicle based on the generated second driving route, since it is the output of the neural network.); and
wherein generating the AI model comprises:
generating a plurality of virtual driving route data by reflecting characteristics of actual driving route data in a reference driving route data, wherein at least some of the generated virtual driving route data have a rounded shape in a corner section and a meandering shape in a straight section, and perform machine learning to train the AI model based on the actual driving route data and the virtual driving route data (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data and dead reckoning data is used to train the neural network. This is machine learning. The position data is actual driving route data. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training. If the vehicle were driven over roads in a typical manner for any length of time, data from corner and straight road sections would be gathered. Corner sections are inherently driven in a rounded manner in a car, which would be reflected in the gathered position data. Human drivers inherently have an element of meandering, i.e., left and right drift, while driving in straight sections due to imprecision in steering and lateral drift of the automobile. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training. This would cause the neural network’s output, i.e. the virtual route data, to reflect the shape of the gathered data, which in this case would include rounded shapes in corner sections and a meandering shape in straight sections.).
Yao does not explicitly disclose detecting a parking position, wherein the first driving route is configured as a plurality of unit vectors, determining a final parking position of the vehicle in the indoor parking.
Song, in the same field of endeavor and solving a related problem, discloses detecting a parking position (See [0021], the method uses dead reckoning to estimate the position of a vehicle in a parking lot. This is a parking position.) determining a final parking position of the vehicle in the indoor parking (See Abstract, the method estimates the driving route of the vehicle in the indoor parking lot using dead reckoning, an accelerometer, and a gyroscope. See [0021], the method estimates the position of a vehicle in a parking lot. This is a parking position. See [0005], the method updates the estimate using map matching. This is a final parking position.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include estimating the vehicle position in an indoor parking lot and determining the parking position of Song. One of ordinary skill in the art would have been motivated to make this modification in order to accurately record where a car has been parked, as suggested by Song at [0002]. Yao combined with Song does not explicitly disclose wherein the first driving route is configured as a plurality of unit vectors.
Zhang renders obvious wherein the first driving route is configured as a plurality of unit vectors (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Regarding claim 20, Yao combined with Song and Zhang renders obvious the limitations of claim 11. Yao discloses connecting the second driving route to a driving route of the vehicle at the time of arrival to determine the final position of the vehicle (See page 5 paragraph 1, the control unit uses the real-time GPS position when it is available, i.e. a driving route, and the dead reckoning position corrected by the neural network, i.e. the second driving route, when the GPS location is not available, i.e. at the time of arrival. This is sequentially connecting the routes. This continues while the vehicle is in operation and therefore determines the final position of the vehicle.). Song, in the same field of endeavor and solving a related problem, renders obvious at the indoor parking lot (See Abstract, the method estimates the position of a vehicle in an indoor parking lot.); and parking position (See [0021], the method estimates the position of a vehicle in a parking lot. This is a parking position. See [0005], the method updates the estimate using map matching. This is a final parking position.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include estimating the vehicle position in an indoor parking lot and determining the parking position of Song. One of ordinary skill in the art would have been motivated to make this modification in order to accurately record where a car has been parked, as suggested by Song at [0002].
Claims 5-7 and 15-17 are rejected under 35 U.S.C. 103 as being obvious over Yao, Song, Zhang, and NPL document “OPLU”, hereinafter “Chernodub”.
Regarding claim 5 Yao combined with Song and Zhang renders obvious the limitations of claim 4. Yao discloses the actual driving route data and the virtual driving route data include a vector reflecting a driving speed and direction of the vehicle (See page 3 paragraph 6, the actual data, which is the data comprising GPS data, includes a velocity vector. See page 4 paragraph 17-page 5 paragraph 1, when the GPS data is not available, the position estimated by dead reckoning and neural network is combined with the available velocity data. This data is therefore the virtual driving route data and also includes a vector reflecting driving speed and direction.). Zhang renders obvious the actual driving route data includes a unit vector (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Chernodub renders obvious the virtual driving route data includes a unit vector (See Abstract, the activation function allows for norm preservation on any layers of the neural network, and therefore norm preservation from the output of the entire output. This means that unit vector inputs will remain unit vector outputs. When the input data is normalized to unit vectors, as would occur from the modification described above, this would mean that the network’s output, i.e. the virtual data, would include unit vectors.). 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 system for using a neural network to correct dead reckoning error in a vehicle, where the neural network is trained on unit vector normalized data, disclosed by Yao, Song, and Zhang to include constraining the neural network to also output unit vectors of Chernodub. One of ordinary skill in the art would have been motivated to make this modification in order to improve the training process, as suggested by Chernodub at Abstract.
Regarding claim 6, Yao combined with Song, Zhang, and Chernodub renders obvious the limitations of claim 5. Yao renders obvious a vector space configuring the reference driving route data includes a straight section and a corner section (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data, i.e. reference driving route data, and dead reckoning data gathered while the vehicle is in operation is used to train the neural network. If the vehicle were driven over roads in a typical manner for any length of time, data from straight and corner sections would be gathered.), generating the plurality of virtual driving route data includes generating the plurality of virtual driving route data by reflecting a characteristic of the unit vector configuring the actual driving route data in the corner section (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data, i.e. actual driving route data, and dead reckoning data gathered while the vehicle is in operation is used to train the neural network. If the vehicle were driven over roads in a typical manner for any length of time, data from a corner section would be gathered. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include gathering training data from straight and corner sections of the road. One of ordinary skill in the art would have been motivated to make this modification because training data will allow improved accuracy of the system, as suggested by Yao at page 5 paragraph 4. Zhang renders obvious unit vectors (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Regarding claim 7, Yao combined with Song, Zhang, and Chernodub renders obvious the limitations of claim 6. Yao renders obvious generating the plurality of virtual driving route data includes generating the plurality of virtual driving route data by reflecting a characteristic of the vector configuring the actual driving route data in the straight section. (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data, i.e. reference driving route data, and dead reckoning data gathered while the vehicle is in operation is used to train the neural network. If the vehicle were driven over roads in a typical manner for any length of time, data from straight and corner sections would be gathered. . To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include gathering training data from straight sections of the road. One of ordinary skill in the art would have been motivated to make this modification because training data will allow improved accuracy of the system, as suggested by Yao at page 5 paragraph 4. Zhang renders obvious unit vector (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Regarding claim 15 Yao combined with Song and Zhang renders obvious the limitations of claim 11. Yao discloses the actual driving route data and the virtual driving route data are formed of a vector reflecting a driving speed and direction of the vehicle (See page 3 paragraph 6, the actual data, which is the data comprising GPS data, includes a velocity vector. See page 4 paragraph 17-page 5 paragraph 1, when the GPS data is not available, the position estimated by dead reckoning and neural network is combined with the available velocity data. This data is therefore the virtual driving route data and also includes a vector reflecting driving speed and direction.). Zhang renders obvious a unit vector (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Chernodub renders obvious the virtual driving route data is formed of a unit vector (See Abstract, the activation function allows for norm preservation on any layers of the neural network, and therefore norm preservation from the output of the entire output. This means that unit vector inputs will remain unit vector outputs. When the input data is normalized to unit vectors, as would occur from the modification described above, this would mean that the network’s output, i.e. the virtual data, would include unit vectors.). 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 system for using a neural network to correct dead reckoning error in a vehicle, where the neural network is trained on unit vector normalized data, disclosed by Yao, Song, and Zhang, to include constraining the neural network to also output unit vectors of Chernodub. One of ordinary skill in the art would have been motivated to make this modification in order to improve the training process, as suggested by Chernodub at Abstract.
Regarding claim 16, Yao combined with Song, Zhang, and Chernodub renders obvious the limitations of claim 15. Yao renders obvious a vector space configuring the reference driving route data includes a straight section and a corner section (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data, i.e. reference driving route data, and dead reckoning data gathered while the vehicle is in operation is used to train the neural network. If the vehicle were driven over roads in a typical manner for any length of time, data from straight and corner sections would be gathered.), generating the virtual driving route data includes generating the plurality of virtual driving route data by reflecting a characteristic of the unit vector configuring the actual driving route data in the corner section (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data, i.e. actual driving route data, and dead reckoning data gathered while the vehicle is in operation is used to train the neural network. If the vehicle were driven over roads in a typical manner for any length of time, data from a corner section would be gathered. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include gathering training data from straight and corner sections of the road. One of ordinary skill in the art would have been motivated to make this modification because training data will allow improved accuracy of the system, as suggested by Yao at page 5 paragraph 4. Zhang renders obvious unit vectors (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Regarding claim 17, Yao combined with Song, Zhang, and Chernodub renders obvious the limitations of claim 16. Yao renders obvious generating the virtual driving route data further includes generating the plurality of virtual driving route data by reflecting a characteristic of the vector configuring the actual driving route data in the straight section. (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data, i.e. reference driving route data, and dead reckoning data gathered while the vehicle is in operation is used to train the neural network. If the vehicle were driven over roads in a typical manner for any length of time, data from straight and corner sections would be gathered. . To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include gathering training data from straight sections of the road. One of ordinary skill in the art would have been motivated to make this modification because training data will allow improved accuracy of the system, as suggested by Yao at page 5 paragraph 4. Zhang renders obvious unit vector (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being obvious over Yao, Song, and Zhang, in view of NPL document “Destination prediction”, hereinafter “Xue”.
Regarding claim 9, Yao combined with Song and Zhang renders obvious the limitations of claim 1. Yao combined with Song and Zhang does not explicitly disclose generating the second driving route includes generating the second driving route in a grid form. Xue, in the same field of endeavor and solving related problem, renders obvious generating the second driving route includes generating the second driving route in a grid form (See page 256, column 2, paragraph 2. Vehicle trajectories are represented in grid format.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include generating data in grid format as disclosed by Xue. One of ordinary skill in the art would have been motivated to make this modification because the grid format allows more accurate inference about destination prediction and inference, as suggested by Xue at page 254, column 2, paragraph 2-page 255, column 1, paragraph 1.
Regarding claim 19, Yao combined with Song and Zhang renders obvious the limitations of claim 11. Yao combined with Song and Zhang does not explicitly disclose generating the second driving route includes generating the second driving route in a grid form. Xue, in the same field of endeavor and solving related problem, renders obvious generating the second driving route includes generating the second driving route in a grid form (See page 256, column 2, paragraph 2. Vehicle trajectories are represented in grid format.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include generating data in grid format as disclosed by Xue. One of ordinary skill in the art would have been motivated to make this modification because the grid format allows more accurate inference about destination prediction and inference, as suggested by Xue at page 254, column 2, paragraph 2-page 255, column 1, paragraph 1.
Response to Arguments
(A) Applicant argues “Rejection under 35 U S.C. 112(b)
Claims 4-8 and 14-18 were rejected under 35 USC12(b) as being indefinite, and in
response, claims 4 and 14 have been canceled without prejudice, whereas claims 6 and 16 have been amended to cancel the language "in a shape of a polygon," thereby obviating the rejection.”
As to (A)¸ Examiner does not find the argument persuasive. As explained in the 112b section above, a vector space is either incapable of representing a driving route or inherently comprises all possible driving routes, in which case claims 6 and 16 would provide no additional limitations to claims 5 and 15.
(B) Applicant argues “Rejection under 35 U S.C. 101
Claims 1-20 were rejected under 35 USC 101 as being "directed to an abstract idea without significantly more." This rejection is respectively traversed.
As stated on page 10 of the "October 2019 Update: Subject Matter Eligibility," the "mere recitation of a judicial exception does not mean that the claim is 'directed to' that judicial exception ... if the claim as a whole 'integrates the recited judicial exception into a practical application of that exception'."
In the subject application, the specification provides an "improvement," and "the claim itself reflects the disclosed improvement" (see page 12 of the "October 2019 Update: Subject Matter Eligibility").
Under Section II (Prong Two, Step 2B), it is necessary to "evaluate whether the judicial exception is integrated into a practical application" (see page 18 of the "2019 Revised Patent Subject Matter Eligibility Guidance"). In this case, independent claim 1 amounts to significantly more than an abstract idea itself, at least because it recites improvements to the functioning of a
computer, i.e., a processor is configured to:"generate a first driving route based on dead reckoning (DR) using speed and direction information of the vehicle, wherein the first driving route is configured as a plurality of unit vectors,""generate an artificial intelligence (AI) model for error correction by performing machine learning,""generate a second driving route by inputting the generated first driving route into the generated AI model, the second driving route being generated by sequentially correcting an error in each of the plurality of unit vectors in chronological order through the AI model," and "determine a final parking position of the vehicle in the indoor parking lot based on the generated second driving route" (see independent claim 1; see also independent claim 11). In other words, the apparatus and method according to the claimed invention are configured to generate a first driving route formed of a plurality of unit vectors, and subsequently generate a second driving route by inputting the unit vectors into an AI model, in order the determine a final parking position of a vehicle in an indoor parking lot.
In addition to improvements to the functioning of a computer, Prong Two (Step 2B) is
satisfied at least because the additional element/combination of elements is "linked to a particular technological environment," i.e., determining a final parking position of a vehicle in an indoor parking lot by utilizing a plurality of unit vectors (representing speed and direction information) of a "first" driving route and error correcting the same using an AI model (see, e.g., page 18, lines 4-8 of the specification). Therefore, the claim amendments and above explanation address and overcome the rejection under 35 USC 101.
In summary, the claimed invention cannot be considered "mental processes" (see page 9, last paragraph of the Office Action of 07/14/2025), at least because the claimed invention requires utilizing a plurality of unit vectors (representing speed and direction information), generating a "second" driving route by inputting the same into an AI model to sequentially correct an error in each of the unit vectors in chronological order, and determining a final parking position of a vehicle in an indoor parking lot based on the error-corrected "second" driving route, which cannot be considered "mental processes."”
As to (B)¸Examiner does not find the argument persuasive. The specification of unit vectors does not render the process impossible in the human mind. The unit vector is a representation of the quantity of interest in computer memory, similarly to how all computer-executable programs must represent all data of interest as sequences of bits in memory. The inability of a human to rapidly perform complex binary arithmetic does not mean that a computer program inherently does not describe a mental process execute on a computer. The same argument shows that representation of the route as unit vectors does not make the route estimation by dead reckoning impossible using the human mind or, for example, using a calculator and rough estimates. Implementation of a mental process on a computer is not enough to show improvement to technological functioning, at least because the remaining specifically mentioned limitations are well-known, routine, and conventional.
(C) Applicant argues “Rejections under 35 U.S.C. X103
Claims 1, 10, 11, and 20 were rejected under 35 USC 103 as being unpatentable over
Chinese Patent 108871336 to Yao et al. ("Yao") in view of Chinese Patent 103453913 to Song et al. ("Song"). The remaining dependent claims were rejected based on combinations including at least the Yao and Song references. These rejections are respectfully traversed.
Regarding the rejection of independent claims 1 and 11 over the proposed combination of Yao in view of Song, the proposed combination does not teach or suggest an "apparatus for detecting a parking position of a vehicle," including:
a processor; and
a memory storing executable instructions that, when executed by the processor,
cause the processor to:
generate a first driving route based on dead reckoning (DR) using speed and direction information of the vehicle, wherein the first driving route is
configured as a plurality of unit vectors;
generate an artificial intelligence (AI) model for error correction by performing machine learning;
generate a second driving route by inputting the generated first driving route into the generated AI model, the second driving route being generated by
sequentially correcting an error in each of the plurality of unit vectors in
chronological order through the AI model;
determine a final parking position of the vehicle in the indoor parking lot based on the generated second driving route,
wherein in generating the AI model, the processor is configured to: generate a plurality of virtual driving route data by reflecting characteristics of actual driving route data in a reference driving route data,
wherein at least some of the generated virtual driving route data have a rounded
shape in a corner section and a meandering shape in a straight section, and perform machine learning to train the AI model based on the actual driving route data and the virtual driving route data.
See independent claim 1 (see also independent claim 11).
Referring to the English-language abstract of Yao, it is described that when a "GPS receiver does not have position data output," then a dead reckoning model is used to predict position errors by using a Kalman filter.
Referring to the English-language abstract of Song, it is described that accurate
positioning of a vehicle in a parking lot is obtained by using a "DR ... module and an indoor parking lot electronic map matching module," while incorporating use of a Kalman filter.
On page 19 of the Office Action of 07/14/2025, it was acknowledged that the Yao reference does not teach or suggest the use of unit vectors.
The NPL reference "Understand Data Normalization in Machine Learning" provides a mathematical equation for unit vector normalization.
However, even if the above NPL reference is somehow combined with Yao in view of Song, it is apparent that the proposed combination does not teach or suggest generating a first driving route based on dead reckoning using speed and direction information of a vehicle by utilizing a plurality of unit vectors (representing speed and direction information), generating a "second" driving route by inputting the same into an AI model to sequentially correct an error in each of the unit vectors in chronological order, and determining a final parking position of a vehicle in an indoor parking lot based on the error-corrected "second" driving route.
Further, there is no teaching or suggestion in the proposed combination of Yao in view of Song (and/or the NPL reference) that generating an AI model includes generating a plurality of virtual driving route data by "reflecting characteristics of actual driving route data in a reference driving route data," where "at least some of the generated virtual driving route data have a rounded shape in a corner section and a meandering shape in a straight section," and performing
"machine learning to train the AI model based on the actual driving route data and the virtual driving route data" (see independent claim 1; see also independent claim 11). “
As to (C), Examiner does not find the argument persuasive. Yao discloses An apparatus for detecting a position of a vehicle (See Abstract, the position estimating system uses sensor fusion and angle data to obtain real-time vehicle position), the apparatus comprising:
a processor (See page 2 paragraph 5-page 3 paragraph 1, the system uses received GPS information and vehicle information in a Kalman filter, which is a sequence of mathematical operations. This necessarily takes place on a system comprising a processor with associated memory.); and
a memory storing executable instructions (See page 2 paragraph 5-page 3 paragraph 1, the system uses received GPS information and vehicle information in a Kalman filter, which is a sequence of mathematical operations. This necessarily takes place on a system comprising a processor with associated memory storing the corresponding instructions.) that, when executed by the processor, cause the processor to:
generate a first driving route based on dead reckoning (DR) using speed and a direction information of the vehicle (See page 5 paragraph 7, dead reckoning estimates the position of the vehicle. The sequence of estimated positions of the vehicle estimated by dead reckoning is a first driving route. See page 3 paragraph 8, the plurality of sensors including a speedometer and gyroscope that provides angle, i.e. direction data, are used to perform dead reckoning.);
generate an artificial intelligence (Al) model for error correction by performing machine learning (See page 4 paragraph 5-7, a neural network is trained, i.e. generated, to correct the dead reckoning error.).
generate a second driving route by inputting the generated first driving route into the generated Al model, the second driving route being generated by sequentially correcting an error in each of the plurality of unit vectors in chronological order through the Al model (See page 5 paragraph 7, the dead reckoning system estimates the position of the vehicle. The uncorrected position estimates are the first driving route. See page 5 paragraph 11, the model prediction unit corrects the position estimated by dead reckoning using a neural network, with the neural network taking the position estimated by dead reckoning as input. The sequence of estimated positions of the vehicle estimated by the model prediction unit is a second driving route. See page 7 paragraphs 14-15, the system is used to obtain real-time vehicle estimates. This means that the vectors that make up the first driving route are corrected to produce the second driving route as they are collected, which is sequentially.);
determine a final position of the vehicle based on the generated second driving route (See page 4 paragraph 17-page 5 paragraph 1, an improved estimate of the vehicle’s location is outputted. The last outputted position is a final position of the vehicle based on the generated second driving route, since it is the output of the neural network.); and
wherein in generating the Al model, the processor is configured to:
generate a plurality of virtual driving route data by reflecting characteristics of actual driving route data in a reference driving route data, wherein at least some of the generated virtual driving route data have a rounded shape in a corner section and a meandering shape in a straight section, and perform machine learning to train the AI model based on the actual driving route data and the virtual driving route data (See page 2 paragraph 7-page 3 paragraph 1, vehicle position data and dead reckoning data is used to train the neural network. This is machine learning. The position data is actual driving route data. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training. If the vehicle were driven over roads in a typical manner for any length of time, data from corner and straight road sections would be gathered. Corner sections are inherently driven in a rounded manner in a car, which would be reflected in the gathered position data. Human drivers inherently have an element of meandering, i.e., left and right drift, while driving in straight sections due to imprecision in steering and lateral drift of the automobile. To train a neural network to correct the error between dead reckoning estimated position and the actual position, the dead reckoning position must be input into the neural network. The positions outputted from the neural network are virtual driving route data because they are produced by the neural network, and not directly gathered from the vehicles. The training process of a neural network reduces the discrepancy between the outputted values and the training data, in this case the actual position data. This means that the training process causes the neural network’s virtual driving route data to reflect the characteristics of the actual driving route data. The trained neural network has the same property when it is used for inference after training. This would cause the neural network’s output, i.e. the virtual route data, to reflect the shape of the gathered data, which in this case would include rounded shapes in corner sections and a meandering shape in straight sections.).
Yao does not explicitly disclose detecting a parking position, wherein the first driving route is configured as a plurality of unit vectors, determining a final parking position of the vehicle in the indoor parking.
Song, in the same field of endeavor and solving a related problem, discloses detecting a parking position (See [0021], the method uses dead reckoning to estimate the position of a vehicle in a parking lot. This is a parking position.) determining a final parking position of the vehicle in the indoor parking (See Abstract, the method estimates the driving route of the vehicle in the indoor parking lot using dead reckoning, an accelerometer, and a gyroscope. See [0021], the method estimates the position of a vehicle in a parking lot. This is a parking position. See [0005], the method updates the estimate using map matching. This is a final parking position.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include estimating the vehicle position in an indoor parking lot and determining the parking position of Song. One of ordinary skill in the art would have been motivated to make this modification in order to accurately record where a car has been parked, as suggested by Song at [0002]. Yao combined with Song does not explicitly disclose wherein the first driving route is configured as a plurality of unit vectors.
Zhang, referred to previously as “Understand Data Normalization in Machine Learning”, renders obvious wherein the first driving route is configured as a plurality of unit vectors (See page 2 paragraph 5-page 3 paragraph 1, use of unit vectors is a known method of normalization in machine learning. See page 13 paragraphs 3-4, use of normalization has been empirically observed to improve accuracy and reduce training time in neural networks. Examiner notes that neural networks trained on normalized data normalize input to the network before inference, otherwise the network’s use in inference does not match its training setting.). 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 system for using a neural network to correct dead reckoning error in a vehicle disclosed by Yao to include normalizing the route information to unit vectors as disclosed by Zhang, including normalization of the vectors during inference, i.e. during operation of the vehicle. One of ordinary skill in the art would have been motivated to make this modification in order to improve accuracy of the neural network and reduce training time, as suggested by Zhang at page 13 paragraphs 3-4.
Additional Relevant Art
The prior art made of record and not relied upon is considered pertinent to the applicant’s disclosure and may be found on the accompanying PTO-892 Notice of References Cited:
US 20190063947 A1 which relates to use of dead reckoning to establish the parking position of a vehicle in an indoor parking facility.
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 AUSTIN ROBERT CHENNAULT whose telephone number is (571)272-4606. The examiner can normally be reached Monday - Friday 9:00am - 5:00pm 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, Hitesh Patel can be reached at (571) 270-5442. 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.
/AUSTIN ROBERT CHENNAULT/Examiner, Art Unit 3667
/ANSHUL SOOD/Primary Examiner, Art Unit 3667