DETAILED ACTION
Drawings
The drawings are objected to under 37 CFR 1.83(a) for failing to show fail to show details of the block diagrams, methods, and flowcharts of Figure 1-3 since the rectangular boxes are depicted without descriptors as to the functional flow as described in the specification. The unlabeled rectangular boxes depicted in the drawings should be provided with descriptive text labels. Any structural detail that is essential for a proper understanding of the disclosed invention should be shown in the drawing. MPEP § 608.02(d). Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. In addition, the drawings should be clarified in view of the specification. The published specification appears to indicate in ¶ 20 that block 102 executes SGOUT which is applied to a GMM and an LPD model. Then, in ¶ 21-22, the specification indicates that the GMM is executed at block 104 and block 106 executes LPD such that it appears the steps have overlapping functions. It is recommended to clearly indicate what the output is at each step and what is performed at each step. There appears to be a similar overlap with respect to FIG. 3 where LPD is (again?) performed at step 202. It is recommended, in addition to labeling the blank boxes, clearly indicating how Fig. 3 interrelates with FIG. 2.
Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. The figure or figure number of an amended drawing should not be labeled as “amended.” If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Specification
The disclosure is objected to because of the following informalities: the published specification ¶ 20 states “Block also entails executing SGOUT” such that the specification fails to disclose which block includes SGOUT as recited in claim 10. Appropriate correction is required.
Claim Rejections - 35 USC § 112
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 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
With respect to claims 1 and 11, the metes and bounds of what is and is not included in the limitation “determining, using a lane-point detection (LPD) model, a second set of lane
line points” under a broadest reasonable interpretation since the specification indicates determination of the second set and first set of lane line points could occur in the same step and simultaneously (block 102, ¶ 20). It is unclear what attributes of the “second set” of lane line points generated at step 106 differentiates it from the “first set” of lane line points generated at step 104, since the specification does not provide a limiting definition on “second set” lane line points and the fact that these points are generated using lane point detection does not differentiate them from the first set of lane line points. Both are indicated as “relating to the lane lines of the roadway within the predetermined region”. It is recommended to provide a differentiating attribute of what these second set of points are required to possess that is different than the first set in the claim language. In addition, it is unclear if the first points are included within the second points, i.e., are the second points a filtered or otherwise transformed version of the first points? Or, are these a completely different set of points unrelated to the first set? The specification appears to imply (¶¶ 19-20) that the lane line points may be the received telemetry data, but this point is not made clear. Is the received telemetry data transformed into the first and second lane line points or are these unrelated data points? It is recommended to provide proper antecedent characterization if the lane line points are generated from the telemetry data or are the telemetry data. To the extent the first and second lane line points are actually the telemetry points, but organized, analyzed, used to render statistics, it is recommended to indicate in the claim language that these points are not generated by the GMM and/or LPD, but that the telemetry data points are analyzed or sorted by these functions.
Furthermore, does the output of step 104 provide the generated first points to step 106 to further refine the first points into the second points? Are these unrelated steps? Both steps appear to determine peaks to detect lane lines (Spec. ¶¶ 21-22). Accordingly, it is also unclear how lane lines are determined “using a lane-point clustering (LPC) model . . . based on the first set of lane line points and the second set of lane line points”. The specification merely treats this step as a black box without describing how the inputs are translated to outputs. Spec. ¶ 43 appears to be the only description of this process and fails to specifically account for both lane point inputs and explain how both of these inputs are combined to produce an output based on them.
Claims 1 and 11 are rejected for an additional reason. The limitation “determining, using a Gaussian Mixture Model (GMM), a first set of lane line points” is unclear and indefinite in view of the specification and remaining claim language, i.e., including claim 10, and similarly claim 20, reciting “executing a synthetic gaussian oversampling undersampling technique (SGOUT). The specification indicates “SGOUT is used to generate the graph of the perpendicular distance versus the density of the map matched telemetry” (¶ 20). Claim 9 requires “executing the GMM comprises . . . generating a graph of the perpendicular distance versus a density of the map-matched telemetry data”. Accordingly, it is unclear which Gaussian technique generates the graph of the perpendicular distance versus a density. The specification further confuses the issue by stating SGOUT can occur at first step 102, can be applied to the GMM or LPD model (¶ 20) and that the GMM output graph “includes the output of a synthetic gaussian oversampling undersampling technique (SGOUT). Are these separate steps? If so, why do they have the same output and what is the required order of execution? The metes and bounds of when these steps occur and what their precise output is should be clarified in the specification, claims and drawings.
Claims 1 and 11 are rejected for an additional reason. The limitation “the lane data” recites lack of antecedent basis and further introduces a lack of clarity since unlike the telemetry data, no lane data is ever received. Is the lane data included within either telemetry data, map data or some other source?
Claims 2-4, 10, 12-14 and 20 are rejected for reciting unclear antecedent characterization. In each of these claims a step that appears to be a further limitation of a parent step does not refer back to the parent step such that it is unclear if the further limitation is an aspect of a parent step or is a new and separate step. For example, claim 2 recites “further comprising determining a local density map”. The specification appears to indicate the second set of points are generated based on the determined local density map (Spec. ¶ 22) such that it is unclear if local density map determination occurs in claim 1 to produce the second set and claim 2 is referring back to this process, or if claim 2 is requiring an additional local density map determination outside of the process required to generate the second set of llps (SLLP).
Claims 10 and 20 are rejected for at least an additional reason. The metes and bounds of what is and is not included in the term “synthetic gaussian oversampling undersampling technique (SGOUT)” is unclear and indefinite. The term is not provided with a limiting definition and the specification provides a series of actions that may or may not be taken when executing SGOUT in ¶ 20. It is recommended to specify in the claim language which of these steps (distance v. density, aggregating into segments, distribution/ peaks, mode size, upper limit, lower limit, mode shift, Gaussian kernel, mean and standard deviation, generating a new distribution with equal mode density, inputting graph into perpendicular distance versus density map matched telemetry data, etc.) is required when “executing . . . SGOUT” and which are not. SGOUT has been extensively searched and is not a well known term in the art. In addition, several of the above steps appear to overlap with other recited claim limitations such that it is unclear what constitutes SGOUT execution, i.e., claims 2-4 appear to include some but not all of the potential functions associated with the term in the specification.
Claim Rejections - 35 USC § 112(a)
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the enablement requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to enable one skilled in the art to which it pertains, or with which it is most nearly connected, to make and/or use the invention.
With respect to the limitation “determining . . . a first set of lane line points relating to the lane lines of the roadway . . . determining . . . a second set of lane line points relating to the lane lanes of the roadway . . . determining, using a lane-point clustering (LPC) model, a plurality of lane lines of the roadway based on the first set of lane line points and the second set of lane line points”, as recited in claims 1 and 11, the specification fails to enable how a LPC model inputs a first set of points output from a gaussian mixture model and a second set of points output from a LPD model. The specification indicates the LPC itself detects the first and second lane line points (¶ 43) such that it is unclear how the claim requires these are inputs to the LPC rather than generated within the LPC. The specification appears to indicate all steps are performed within black box 110, FIG. 2 (¶ 43). The specification fails to a enable a PHOSITA to make or use an invention that somehow combines the output of the GMM and LPD models as inputs to an LPC model to produce an output described as “the shape, location, and extent of the lane lines of the roadway”, particularly when the specification indicates this is the same output of the GMM constituting the first points, ¶ 21, “block 104 . . . using a . . . GMM . . . line interpolation . . . In other words, the controller 12 then determines the shape, location, and extent of the lanes lines of the roadway”.
To satisfy the enablement requirement of 35 U.S.C. 112(a) or pre-AIA 35 U.S.C. 112,
first paragraph, the specification must teach those skilled in the art how to make and use the full
scope of the claimed invention without "undue experimentation." See, e.g., In re Wright, 999
F.2d 1557, 1561, 27 USPQ2d 1510, 1513 (Fed. Cir. 1993); MPEP 2161.01, III. “The Federal Circuit has stated that "‘[i]t is the specification, not the knowledge of one skilled in the art, that must supply the novel aspects of an invention in order to constitute adequate enablement.’" Auto. Technologies, 501 F.3d at 1283, 84 USPQ2d at 1115 (quoting Genentech, Inc. v. Novo Nordisk A/S, 108 F.3d 1361, 1366, 42 USPQ2d 1001, 1005 (Fed. Cir.1997)). The rule that a specification need not disclose what is well known in the art is "merely a rule of supplementation, not a substitute for a basic enabling disclosure." Genentech, 108 F.3d at 1366. See MPEP 2161.01, III. Furthermore, a rejection under 35 U.S.C. 112(a) for lack of enablement must be made when the specification does not enable the full scope of the claim. See MPEP 2161.01, III. Here, the specification fails to enable the full scope of the limitations
The standard for determining whether the specification meets the enablement requirement was cast in the Supreme Court decision of Minerals Separation v. Hyde, 242 U.S. 261,270 (1916) which posed the question: is the experimentation needed to practice the invention undue or unreasonable? That standard is still the one to be applied. In re Wands, 858 F.2d 731,737 (Fed. Cir. 1988).
Determining enablement is a question of law based on underlying factual findings, In re Vaeck, 947, F.2d 488, 495 (Fed. Cir. 1991). The determination that “undue experimentation” would have been needed to make and/or use the claimed invention is not a single, simple factual determination. Rather it is a conclusion that may be reached by weighing some or all of the following non-exhaustive list of factual considerations:
(A) the breadth of the claims;
(B) the nature of the invention;
(C) the state of the prior art;
(D) the level of one of ordinary skill;
(E) the level of predictability in the art;
(F) the amount of direction provided by the inventor;
(G) the existence of working examples; and
(H) the quantity of experimentation needed to make or use the invention based on the content of the disclosure. Wands, 858 F.2d at 737.
With respect to claims 1 and 11, under the Wands factors (MPEP 2164.01(a) A-H), with respect to the above limitations, there are:
(G) no working examples for the full claim scope, i.e., if or how the outputs of the GMM and LPD are combined or otherwise as inputs to the LPC.
(F) no direction is clearly provided as noted above and in the 112(b) rejection. What specifically is output from the GMM and LPD is not clear, i.e., if the output of the GMM is used as an input to the LPD or not, such that only the output of the LPD is entered into the LPC, the specification further indicates the output of the GMM and the LPC can be the same. It is generally unclear what each step is required to perform as noted in the 112(b) rejection.
(B) a highly complex nature of the invention.
(A) the breadth of the claims is such that it presents overlapping or conflicting steps as noted above in the 112(b) rejection and is further exacerbated by the drawing and specification objections noted above, i.e., Spec. ¶¶20-24 and 41-42, i.e., each step is discussed as having a multitude of potential sub-steps, where the sub-steps may occur in multiple different sub-steps, including clustering, such that it is unknown what the inputs the LPC are required to be and what operations the LPC performs to determine a plurality of lane lines, i.e., Spec. ¶ 20 indicates lane lines can be determined at the GMM step, such that a PHOSITA would not be able to make or use an invention that requires “determining, using a lane-point clustering (LPC) model, a plurality of lane lines of the roadway based on the first set of lane line points and the second set of lane line points” given the guidance in the specification without undue experimentation. For the LPC step, the specification fails to discuss how the GMM first line points are input and combined to achieve an output, i.e., rather confusingly, ¶ 43 states the LPC itself detects the first and second set of lane line points.
Therefore, based upon the disclosure of Applicant, one of ordinary skill in the art would not be able to make or use the invention without undue experimentation. The Examiner therefore concludes that the specification does not enable, in combination with the other limitations of the claims, how to make or use the claimed invention. Claims dependent thereon are rejected for the same reasons.
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 1-4, 9, 11-14 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. 20190325739 to Dorum et al. (Dorum) in view of U.S. 20200393265 to Piao et al. (Piao)
With respect to claims 1 and 11, as best understood in view of 112(b) rejection, specification objection and drawing objections above, Dorum discloses a method for estimating lane lines, comprising:
(¶ 63 using probe data in order to estimate lane properties; 83 obtain accurate lane-level geometry)
receiving, by a controller, telemetry data, wherein the telemetry data includes vehicle positional data, the vehicle positional data includes GPS data, [and the lane data includes information about lane lines of a roadway]1 within a predetermined region;
(¶ 65 FIG. 6A illustrates an example of a segment of road 100 including three lanes 102, 104, and 106 separated by lane lines 108. Probe data points 110 are received as raw location points, such as measured GPS location, and are distributed across the roadway 100; lane: ¶ 55 vehicle lane pattern may comprise information regarding the number of lanes along the road segment, a lane identifier for each lane of the road segment, lane center geometry, paint stripe geometry, locations where lanes are added or begin to end/taper; 58; 79 lane separation distance constraint; 89 width of a lane; 76 number of lanes is known; 83 road segment centerline defined in the map data
receiving, by the controller, map data, wherein the map data includes road-network topology data, and the road-network topology data includes a map with information about the roadway within the predetermined region;
(¶ 51 receive digital map from apparatus 10 or provide probe data to apparatus 10; 57; matching the telemetry data with the map data to generate map-matched telemetry data; 83 road segment centerline (defined in the map data))
(¶ 7 for each of the probe data points, determine a location and a road segment corresponding to the location may include processing circuitry configured to: map-match the probe data points to the first road segment; 63 creating a one-dimensional cross-section histogram of a road segment or portion thereof where the probe location points are map-matched to the road center and binned according to their measured distance from the road center map geometry; 77; 83 probe data points that are map-matched to a road segment are binned to a two-dimensional grid that is overlaid on the road segment; ¶ 67 “probe data points are map-matched to the road center”)
determining, using a Gaussian Mixture Model (GMM), a first set of lane line points relating to the lane lines of the roadway within the predetermined region;
(FIG. 7, ¶¶ 65 measured histogram for the raw probe data 110 is illustrated in FIG. 7 . . . Gaussian distribution of probe data about a centerline of the road segment . . . applying the Maximum Entropy Method, as detailed further below, results in the histogram illustrated in FIG. 8, where the number of significant peaks in the resultant histogram corresponds to the number of lanes in the data, and the peak location correlates with the location offset distances of the lanes with respect to the road center line to establish lane locations . . . resultant data from the Maximum Entropy Method is overlaid on the road segment 100 illustrated in FIG. 6B, with lines 112, 114, and 116 representing the significant peaks in the histogram of FIG. 8 correlating with lanes 102, 104, and 106, respectively; 66-69 “gaussian model”, i.e., FIG. 8 with multiple gaussian peaks at lane line intervals)
determining, using a lane-point detection (LPD) model2, a second set of lane line points relating to the lane lanes of the roadway within the predetermined region; and
(i.e., including density distribution and gradient of local density map discussed in Spec. ¶ 22: ¶¶ 66 divided into a pixel grid where the intensity of each pixel is proportional to the number of probe data points whose reported location (e.g., the location data of the probe data point) corresponds with the respective grid cell/pixel . . . the two-dimensional implementation is that detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes; 79-83 map cross sections bins/ cell grids for each road segment or link generated . . . subdividing road segment into multiple sections to detect lane changes . . . FIGs 9A and 9B . . . grid cell histogram Lane center locations may correspond to the cell peaks (or pixels with high intensity) and can be identified with robust statistics or image processing techniques; 83 number of probe data points that fall within that bin/cell on the road segment. This produces a more accurate distribution of probe data along the road segment . . . techniques may be applied as described above to obtain accurate lane-level geometry and traffic data; 86 spatiotemporal cell-density image may be generated from the probe data with a dimension of the image in the time domain representing different periods of time)
determining, using a lane-point clustering (LPC) model, a plurality of lane lines of the roadway based on the first set of lane line points and the second set of lane line points.
(¶ 87 “Once lane-level geometry and traffic information/ data for one or more lanes of the road segment are determined . . . an updated map tile for replacing a map tile in a digital map database or geographic database 21 is generated; 89; 450, FIG. 11; 550, FIG. 12 and corresponding descriptions wherein clustering is used to determine a plurality of lane lines in the roadway, i.e., ¶ 80 “Mean Shift technique may be used to cluster the peaks”; 83-86 probe data points will cluster at the tum-offs and through a turn maneuver, which would be discerned through the two-dimensional grid overlay of the road segment. The two-dimensional embodiment may also be capable of establishing merge lane start/finish and other lane-level details not necessarily discernable through a one-dimensional approach)
With respect to the limitation “and the lane data includes information about lane lines of a roadway” noted in italics above, as discussed in the 112(b) rejection, it is unclear if the claim is requiring the lane data to be received in the telemetry data due to the lack of antecedent basis for “the lane data”, although the lane data recitation is grouped with the telemetry data paragraph rather than the map data paragraph and the specification indicates lane data may be included in the telemetry data.
Dorum fails to explicitly disclose the telemetry data includes lane line detection data. Piao is from the same field of endeavor since Piao also discloses detecting lane lines for high definition maps (title) including gathering lane line detection data included within telemetry data of probe vehicles (FIG. 10 and corresponding description) that is further used to generate a set of lane line point data and lane lines of a roadway using a lane point detection model and a lane point clustering model (Fig. 12B-17 and corresponding descriptions) wherein Piao discloses that telemetry data, i.e., received from vehicles 150, FIG. 1 includes both GPS data and detected information about lane lines including lane line points that is further used to perform map matching and lane line determination on the basis of the lane line points analyzed by a lane point clustering model
(¶¶ 7 receiving a set of one or more lane line points each representing a location on a lane line in an HD map; 130-132 “FIG. 10 illustrates example components used in an example lane line creation process. The camera 2905 may capture a camera image 2910 and may identify 2D points 2915 of those images in a 2D plane 2920. The identified 2D points with high enough center line probabilities may be mapped to lane line points 2925 . . . Each lane line segment may be encompassed by and identified by a lane line segment center-line 2945 which may include two or more lane line points 2925; 133 lane line data provided as additional layer on HD map system; 138-151 lane line point analysis and clustering; 200; 63 online HD map system 110 may be configured to receive sensor data that may be captured by sensors of the vehicles 150 and combine data received from the vehicles 150 to generate and maintain HD maps; 68-69; 76; creation of lane element graph FIG. 22)
Accordingly, it would have been obvious to one of ordinary skill in the art at the time of effective filing date to implement the teachings of Piao in the system of Dorum such that the telemetry data of Dorum includes detected lane line data in order to provide an improved lane point detection model and a more accurate determination of lane lines on the roadway thereby improving HD maps used by vehicles to perform navigation, i.e., lane location with precision of 30 cm or better (Piao ¶110)
(Piao, ¶ 103 map creation module 410 may be configured to create HD map data of HD maps from the sensor data collected from several vehicles 150 that are driving along various routes . . . lane information may have changed as a result of construction in a region, and the map update module 420 may be configured to update the HD maps and corresponding HD map data accordingly . . . lane element graph module 470 may be configured to generate lane element graphs (i.e., a connected network of lane elements) to allow navigation of autonomous vehicles through a mapped area; 122 HD map system 100 may store a lane-centric representation of data that may represent the relationship of the lane to the feature so that the vehicle 710 can efficiently extract the features given a lane)/
With respect to claims 2 and 12, Dorum in view of Piao disclose determining a local density map of the map-matched telemetry data.
(Dorum, ¶¶ 66 divided into a pixel grid where the intensity of each pixel is proportional to the number of probe data points whose reported location (e.g., the location data of the probe data point) corresponds with the respective grid cell/pixel . . . the two-dimensional implementation is that detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes; 79-83 map cross sections bins/ cell grids for each road segment or link generated . . . subdividing road segment into multiple sections to detect lane changes . . . FIGs 9A and 9B . . . grid cell histogram Lane center locations may correspond to the cell peaks (or pixels with high intensity) and can be identified with robust statistics or image processing techniques; 83 number of probe data points that fall within that bin/cell on the road segment. This produces a more accurate distribution of probe data along the road segment . . . techniques may be applied as described above to obtain accurate lane-level geometry and traffic data; 86 spatiotemporal cell-density image may be generated from the probe data with a dimension of the image in the time domain representing different periods of time)
With respect to claims 3 and 13, Dorum in view of Piao disclose
(Dorum, ¶¶ 66 divided into a pixel grid where the intensity of each pixel is proportional to the number of probe data points whose reported location (e.g., the location data of the probe data point) corresponds with the respective grid cell/pixel . . . the two-dimensional implementation is that detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes; 79-83 map cross sections bins/ cell grids for each road segment or link generated . . . subdividing road segment into multiple sections to detect lane changes . . . FIGs 9A and 9B . . . grid cell histogram Lane center locations may correspond to the cell peaks (or pixels with high intensity) and can be identified with robust statistics or image processing techniques; 83 number of probe data points that fall within that bin/cell on the road segment. This produces a more accurate distribution of probe data along the road segment . . . techniques may be applied as described above to obtain accurate lane-level geometry and traffic data; 86 spatiotemporal cell-density image may be generated from the probe data with a dimension of the image in the time domain representing different periods of time)
With respect to claims 4 and 14, Dorum in view of Piao disclose determining a density distribution along the predetermined number of cross-sections using the local density map of the map-matched telemetry data.
(Dorum, ¶¶ 66 divided into a pixel grid where the intensity of each pixel is proportional to the number of probe data points whose reported location (e.g., the location data of the probe data point) corresponds with the respective grid cell/pixel . . . the two-dimensional implementation is that detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes; 79-83 map cross sections bins/ cell grids for each road segment or link generated . . . subdividing road segment into multiple sections to detect lane changes . . . FIGs 9A and 9B . . . grid cell histogram Lane center locations may correspond to the cell peaks (or pixels with high intensity) and can be identified with robust statistics or image processing techniques; 83 number of probe data points that fall within that bin/cell on the road segment. This produces a more accurate distribution of probe data along the road segment . . . techniques may be applied as described above to obtain accurate lane-level geometry and traffic data; 86 spatiotemporal cell-density image may be generated from the probe data with a dimension of the image in the time domain representing different periods of time)
With respect to claims 9 and 19, Dorum in view of Piao disclose the map data is high-definition (HD) map data, and executing the GMM comprises:
determining a perpendicular distance from each of the plurality of telemetry points to a virtual line connecting an adjacent pair of a plurality of HD map points;
(Dorum, ¶ 77 probe data points are assigned a projection distance according to their distance from the road centerline 205 measured at the closest possible distance or perpendicular to the centerline 205)
generating a graph of the perpendicular distance versus a density of the map-matched telemetry data and identifying a plurality of peaks in the graph of the perpendicular distance versus the map-matched telemetry data;
(Dorum, FIG. 7-8 with 8 depicting a plurality of peaks in the graph of the perpendicular distance versus the map-matched telemetry data and corresponding descriptions)
determining a plurality of high-density areas using the plurality of peaks in the graph of the perpendicular distance versus a density of the map-matched telemetry data;
(Dorum, FIG. 11, density histogram 420, determine statistically significant peaks 440 and corresponding description; ¶¶ 65 number of significant peaks in the resultant histogram corresponds to the number of lanes in the data; 77)
determining a plurality of center lines and lane edges of the roadway within a predetermined region.
(Dorum, 450, FIG. 11 and corresponding description; ¶¶ 65 Gaussian distribution of probe data about a centerline of the road segment, number of significant peaks in the resultant histogram corresponds to the number of lanes in the data, peak location correlates with the location offset distances of the lanes with respect to the road center line to establish lane locations; i.e., lane center lines 112, 114, 116 and lane edges defined by bounds of 102, 104, 106 see ¶ 65 “lines 112, 114, and 116 representing the significant peaks in the histogram of FIG. 8 correlating with lanes 102, 104, and 106, respectively)
Claims 5-7 and 15-17 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. 20190325739 to Dorum et al. (Dorum) in view of U.S. 20200393265 to Piao et al. (Piao) and further in view of US 20010056326 to Kimura et al. (Kimura)
With respect to claims 5 and 15, Dorum in view of Piao fail to explicitly disclose a second order derivative filter is applied to the density distribution.
Kimura, from the same field of endeavor, discloses a “method for map matching” (¶ 70) wherein lane lines are detected (¶ 76) by applying a second order derivative filter is applied to a distribution of lane line points in order to detect the attributes of a lane line marker.
(FIG. 13, edge enhancement s12, s13-S14, lane marker detection process S15; ¶¶197-207, i.e., lane marker detector 14 . . . at least two edge pixels . . . are detected when the edge enhancement process is carried out for single scanning line using the Sobel filtering process, Laplacian filtering process, or the like)
Accordingly, it would have been obvious to one of ordinary skill in the art at the time of effective filing date for Dorum in view of Piao to use a second order derivative filter, as taught by Kimura above, and apply it to the density map of the map-matched data in order to provide enhanced boundaries with more clearly delineated edges which improves localizing the properties of lane lines, i.e., second derivative filters like Laplacian filters remove noise and outliers to provide enhanced edge detection (Kimura, ¶ 199 “edge enhancement”)
With respect to claims 6 and 16, Dorum in view of Piao and further in view of Kimura disclose executing the LPD model includes determining the second set of lane line points relating to the lane lines of the roadway within the predetermined region using the gradient of the local density map.
(Dorum, ¶¶ 66 divided into a pixel grid where the intensity of each pixel is proportional to the number of probe data points whose reported location (e.g., the location data of the probe data point) corresponds with the respective grid cell/pixel . . . the two-dimensional implementation is that detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes; 79-83 map cross sections bins/ cell grids for each road segment or link generated . . . subdividing road segment into multiple sections to detect lane changes . . . FIGs 9A and 9B . . . grid cell histogram Lane center locations may correspond to the cell peaks (or pixels with high intensity) and can be identified with robust statistics or image processing techniques; 83 number of probe data points that fall within that bin/cell on the road segment. This produces a more accurate distribution of probe data along the road segment . . . techniques may be applied as described above to obtain accurate lane-level geometry and traffic data; 86 spatiotemporal cell-density image may be generated from the probe data with a dimension of the image in the time domain representing different periods of time)
With respect to claims 7 and 17, Dorum in view of Piao and further in view of Kimura disclose executing the LPD model includes determining a nominal path of the roadway within the predetermined region using the gradient of the local density map.
(Dorum, ¶ 82 “the lane center geometry may be explicitly derived, even for complex shapes such as ramps, round abouts, intersections, etc.”)
(Dorum, ¶¶ 66 divided into a pixel grid where the intensity of each pixel is proportional to the number of probe data points whose reported location (e.g., the location data of the probe data point) corresponds with the respective grid cell/pixel . . . the two-dimensional implementation is that detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes; 79-83 map cross sections bins/ cell grids for each road segment or link generated . . . subdividing road segment into multiple sections to detect lane changes . . . FIGs 9A and 9B . . . grid cell histogram Lane center locations may correspond to the cell peaks (or pixels with high intensity) and can be identified with robust statistics or image processing techniques; 83 number of probe data points that fall within that bin/cell on the road segment. This produces a more accurate distribution of probe data along the road segment . . . techniques may be applied as described above to obtain accurate lane-level geometry and traffic data; 86 spatiotemporal cell-density image may be generated from the probe data with a dimension of the image in the time domain representing different periods of time)
Claims 8 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. 20190325739 to Dorum et al. (Dorum) in view of U.S. 20200393265 to Piao et al. (Piao) and further in view of US 20010056326 to Kimura et al. (Kimura) and further in view of US 20200208992 to Fowe et al. (Fowe)
With respect to claims 8 and 18, Dorum in view of Piao and further in view of Kimura disclose the LPC model includes detecting the lane line points and grouping the lane line points based on changes in lane number or configuration.
(Dorum ¶ 66 “intensity of each pixel is proportional to the number of probe data points whose reported location . . . two-dimensional implementation yields explicit lane center geometry in the form of bright pixels corresponding to each lane center . . . detecting explicit lane center geometry may automatically capture the formation of new lanes and the disappearance of ending lanes, as the two-dimensional implementation gives an accurate representation of the road segment”;
(Dorum ¶ 72 “Where pij is the proportion or probabilities of the total image brightness for the lane center pixel/cell that we wish to identify”;
(Dorum, ¶¶ 79-84 “sub-dividing a road segment into multiple sections can allow the one-dimensional implementation to detect lane changes within a road segment . . . determining lane-level geometries of roadway anomalies and intersections . . . tum-offs of a road segment for businesses, parking garages, or other points-of-interest may be discernable using the two-dimensional approach as probe data points will cluster at the tum-offs and through a turn maneuver, which would be discerned through the two-dimensional grid overlay of the road segment. The two-dimensional embodiment may also be capable of establishing merge lane start/finish and other lane-level details”)
However, Dorum in view of Piao and further in view of Kimura fail to explicitly disclose the LPD model determining attributes that include a confidence level of the second set of lane line points.
Fowe discloses determining attributes for lane line points including a confidence level
(¶ 102 “A confidence metric may also be provided with each selection of a segment. The confidence metric represents a probability that the probe point was matched correctly to a segment. The confidence metric measures the confidence of the accuracy of point map matcher . . . the higher the confidence metric, the higher the likelihood of its accuracy and consequently trust . . . the confidence metric may be stored as an artifact in the geographic database”; 133, 141-142, claims 1-2 and 7-8 gaussian density distribution [Wingdings font/0xE0] probability, mean, variance [Wingdings font/0xE0] confidence based on probability weights)
Accordingly, it would have been obvious to one of ordinary skill in the art at the time of effective filing date to implement a confidence metric in view of the teachings of Fowe, into the system of Dorum in view of Piao and further in view of Kimura in order to improve the accuracy of maps used for vehicle navigation, thereby improving safety and reliability (Fowe, ¶¶ 7, 78, 102-103)
Claims 10 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. 20190325739 to Dorum et al. (Dorum) in view of U.S. 20200393265 to Piao et al. (Piao) and further in view of “CSMOUTE: Combined Synthetic Oversampling and Undersampling Technique for Imbalanced Data Classification”, AGH University of Science and Technology, April 2021 to Koziarski (Koz)
With respect to claims 10 and 20, as best understood in view of the 112(b) rejection above, Dorum in view of Piao fail to explicitly disclose executing a synthetic gaussian oversampling undersampling technique (SGOUT)3 to determine oversampling and undersampling of telemetry data.
Koz, from the same field of endeavor, discloses executing a synthetic gaussian oversampling undersampling technique (SGOUT) to determine oversampling and undersampling technique.
(p. 1, col. 2 “we propose a Combined Synthetic Oversampling and Undersampling Technique (CSMOUTE), which integrates SMOTE oversampling with SMUTE undersampling’; section III “CSMOUTE algorithm”)
Accordingly, it would have been obvious to one of ordinary skill in the art at the time of effective filing date to implement the Oversampling and Undersampling Technique (CSMOUTE) taught by Koz to the telemetry dataset disclosed by Dorum in view of Piao in order to recharacterize the telemetry data groupings to improve accuracy while reducing risk of incorrect classification of borderline cases with respect to both high and low density telemetry groupings (Koz, p. 2, col.1 “methods combining over- and undersampling emerged . . . strengthen the core of a minority class while simultaneously reducing the risk of incorrect classification of borderline instances”) and further because under and oversampling combined can result in better performance than over or under sampling alone (p. 6, col. 1, FIG. 2) and improved performance in datasets with a high number of outliers (Koz, p. 7, col. 1 “the algorithm is particularly well suited for the datasets with a high number of outliers”).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KENNETH J MALKOWSKI whose telephone number is (313)446-4854. The examiner can normally be reached 8:00 AM - 5:00 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Faris Almatrahi can be reached at 313-446-4821. 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.
/KENNETH J MALKOWSKI/Primary Examiner, Art Unit 3667
1 Per the 112(b) rejection above, the claim fails to indicate the source of the lane data such that it is unclear if it must be included within map data, telemetry data or neither.
2 No limiting definition is provided for LPD model but can include something that “determines a local density map”, “divides the map data into a predetermined number of one-meter cross-sections”, provides embedding generation, determines a gradient of the local density map, determine lane line point attributes, determine a confidence level of lane line points, determine changing lane configurations and/or detect how many lanes are included in the roadway (Spec. ¶ 22). Accordingly, under a BRI or the plain meaning of the term, an LPD model is any function that processes or determines any attributes associated with at least two lane line points.
3 As noted in the 112(b) rejection, this term is not known in the industry and no limiting definition is provided for what it is. The term is only defined in terms of what it can be used to generate, i.e., it is something that generates a graph of the perpendicular distance versus the density of the map-matched
telemetry data (Spec. ¶ 20)