PAVEMETRICS SYSTEMS INC v. TETRA TECH INC (2022)

ORDER RE: MOTION FOR SUMMARY JUDGMENT (ECF NO. 108)

FILED PROVISIONALLY UNDER SEAL

Plaintiff Pavemetrics Systems, Inc. moved for partial summary judgment on the ground that its accused Laser Rail Inspection System (“LRAIL”) does not infringe U.S. Patent No. 10,362,293 (“the ’293 Patent”). Mot., ECF No. 108; see also Mem., ECF No. 109. Defendant Tetra Tech, Inc. opposed the motion, Opp'n, ECF No. 125, and Plaintiff replied, Reply, ECF No. 137-1 (filed partially under seal). The Court heard argument on this motion on February 28, 2022. ECF No. 141.

I. BACKGROUND

Plaintiff brought this action against Defendant for a declaratory judgment that its LRAIL product does not infringe the ’293 Patent. Compl. ¶¶ 14–22, ECF No. 1. Defendant filed a counterclaim alleging that Plaintiff's LRAIL product infringes claims 1, 15–16, 19–22, 36–37, and 40–42 of the ’293 Patent. Countercl. ¶¶ 28–37, ECF No. 12. Plaintiff's LRAIL system uses laser triangulation technology to collect data on railway tracks and uses software to process that data and evaluate the conditions of railway track components, such as rails, ties, ballast, and fasteners. Mem. 6 (citing Hébert Decl. ¶¶ 6–9, ECF No. 114-1 (filed under seal)). The Court previously construed several terms in the ’293 Patent. Claim Construction Order, ECF No. 103.

II. LEGAL STANDARD

Summary judgment is appropriate where there is no genuine issue of material fact and the moving party is entitled to judgment as a matter of law. Fed. R. Civ. P. 56(a); Celotex Corp. v. Catrett, 477 U.S. 317, 330, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). A fact is material when, under the governing law, the resolution of that fact might affect the outcome of the case. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). A dispute is genuine “if the evidence is such that a reasonable jury could return a verdict for the nonmoving party.” Id. The burden of establishing the absence of a genuine issue of material fact lies with the moving party, see Celotex, 477 U.S. at 322–23, 106 S.Ct. 2548, and the court must view the facts and draw reasonable inferences in the light most favorable to the nonmoving party, Scott v. Harris, 550 U.S. 372, 378, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007). To meet its burden, “[t]he moving party may produce evidence negating an essential element of the nonmoving party's case, or, after suitable discovery, the moving party may show that the nonmoving party does not have enough evidence of an essential element of its claim or defense to carry its ultimate burden of persuasion at trial.” Nissan Fire & Marine Ins. Co. v. Fritz Cos., 210 F.3d 1099, 1106 (9th Cir. 2000). Once the moving party satisfies its burden, the nonmoving party cannot simply rest on the pleadings or argue that any disagreement or “metaphysical doubt” about a material issue of fact precludes summary judgment. Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 586, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986). There is no genuine issue for trial where the record taken as a whole could not lead a rational trier of fact to find for the nonmoving party. Id. at 587, 106 S.Ct. 1348.

Patent infringement is a question of fact and, therefore, “is amenable to summary judgment when no reasonable factfinder could find that the accused product contains every claim limitation or its equivalent.” Akzo Nobel Coatings, Inc. v. Dow Chem. Co., 811 F.3d 1334, 1339 (Fed. Cir. 2016). The interpretation of a patent claim term that relies solely on evidence intrinsic to the patent is a question of law. Teva Pharms. USA, Inc. v. Sandoz, Inc., 574 U.S. 318, 331, 135 S.Ct. 831, 190 L.Ed.2d 719 (2015).

III. DISCUSSION

Plaintiff argues its LRAIL system does not infringe the ’293 Patent because the LRAIL system does not perform several limitations of the claims in the ’293 Patent. Specifically, Plaintiff argues that the LRAIL system does not generate a “track elevation map” or a “three dimensional elevation map” as defined by the Court, Mem. 14–16, that the LRAIL system does not sequentially and completely process a 3D map like a sliding window, id. at 17, and that the LRAIL system is not “configured,” as defined by the Court, to identify railway track features from certain images, Reply 10. Defendant argues that the LRAIL system produces three different images, a Fake3D Image, a Fake3D Image with Overlay, and a Mixed Image, each of which is a “three dimensional elevation map” as defined by the Court, Opp'n 6–15, that the LRAIL system identifies track features by sequentially and completely processing a 3D map like a sliding window, id. at 15–23, and that the LRAIL system is “configured” to identify railway track features from the Mixed Image, id. at 23.

A. Evidentiary Objections

The parties produce the declarations of two experts to support their infringement positions. Plaintiff presents the declarations of Jean-François Hébert, a co-owner of Pavemetrics Systems, Inc. who has a PhD in Electrical and Computer Engineering with a specialization in neural networks. Hébert Decl. ¶ 1; see also Suppl. Hébert Decl. ¶ 1, ECF No. 137-3 (filed under seal).¹ Defendant presents the declarations of Vassilios Morellas, a Research Professor of Electrical and Computer Engineering at the University of Minnesota, Twin Cities. Morellas Decl. ¶ 1, ECF No. 134-23 (filed under seal); see also Morellas Claim Construction Decl. ¶ 6, ECF No. 86-9.

Plaintiff objects to the Morellas declaration on the grounds that it is conclusory, unsupported by admissible evidence, and contradictory of other evidence in the record. Reply 1. The Court disagrees that the Morellas declaration is conclusory, as it discloses 45 pages of analysis supported by citations of LRAIL source code. See generally Morellas Decl. Morellas's expert opinion does not need to be supported by admissible evidence. See Fed. R. Evid. 703. Finally, while Plaintiff claims Morellas's conclusions contradict other evidence in the record, this is not a basis to find the evidence inadmissible; instead, it goes to the weight of the evidence. i4i Ltd. P'ship v. Microsoft Corp., 598 F.3d 831, 856 (Fed. Cir. 2010).

Defendant's objections to the Hébert declaration fare no better. Defendant also argues that the Hébert declaration is conclusory, Opp'n 2, but both declarations disclose 34 pages of analysis by someone who has close knowledge of the LRAIL technology, see generally Hébert Decl.; Suppl. Hébert Decl. Defendant's argument that the Hébert declaration contradicts other evidence once again goes to the weight of the evidence rather than its admissibility. i4i, 598 F.3d at 856.

B. The ’293 Patent

The ’293 Patent discloses a 3D Track Assessment System and Method that is designed to automate the time-consuming and subjective process of manual railway track inspections. ’293 Patent at 1:1–54. The Court previously defined the three claim limitations the parties analyze in this motion for summary judgment.

The first claim limitation takes two forms, “track elevation map” in claims 1, 15, and 36, ’293 Patent at 29:44, 30:62–63, and 32:61, and “three dimensional elevation map” (“3D map”) in claim 22, ’293 Patent at 31:45. The Court defined these limitations to mean “representation of both the height (elevation) and intensity of reflected laser light (intensity) at each data point within an area across the entire railway track bed.” Claim Construction Order 32. In so defining these limitations, the Court rejected Defendant's argument that the “map” discussed in these terms does not need to be based on both elevation and intensity data. Id. at 28–32. The Court noted that because the specification states that any such map must be “based on” the “three dimensional elevation and intensity data,” that map “must incorporate information from both sets of data.” Id. at 31. The Court, however, did no more. Specifically, the Court did not, as Plaintiff contends, require that each data point of any 3D map include two separate values for elevation and intensity data. See Mem. 13 (“Pavemetrics’ LRAIL software never analyzes a three-dimensional track elevation map that represents both elevation and intensity data.”). Defendant seems to advocate for a different interpretation of this claim limitation wherein a map represents both elevation and intensity data if it blends or merges elevation and intensity data into a single value. Opp'n 8–12. Because the parties differ on whether the claim terms “track elevation map” and “three dimensional elevation map” permit the blending or merging of elevation and intensity data, the Court must further define this limitation to resolve this dispute. See Teva Pharms., 574 U.S. at 331, 135 S.Ct. 831.

A court may use intrinsic evidence such as the prosecution history and the specification language to interpret claims in a patent. Phillips v. AWH Corp., 415 F.3d 1303, 1314 (Fed. Cir. 2005) (en banc). This is because an inventor is free to be their own lexicographer and to define terms and specify the claim scope in a patent. Id. at 1316; see also id. at 1317 (“It is therefore entirely appropriate for a court, when conducting claim construction, to rely heavily on the written description for guidance as to the meaning of the claims.”). The ’293 Patent states that the “track elevation map” is “based on the acquired three dimensional data.” ’293 Patent at 2:58–59, 4:47–48, 29:44–45; accord id. at 31:45–46. The ’293 Patent also states that the “three dimensional elevation map” is “based on the sensed light reflected from the track bed surface,” id. at 4:62–64; accord id. at 5:9–11, and that it is “based on surface elevation and intensity data representative of an area segment of the railway trackbed,” id. at 31:45–47.

In all instances, the specification and claim language state that the 3D maps are “based on” elevation and intensity data. The Federal Circuit has recognized that “based on” has a broader reading than “means” or “is.” Am. Med. Sys., Inc. v. Biolitec, Inc., 618 F.3d 1354, 1361 (Fed. Cir. 2010). By using the language “based on,” the Court concludes the patent claim does not require that the 3D maps encode elevation and intensity data separately at each point. Instead, the 3D maps may encode the elevation and intensity data by blending or merging the values with mathematical formulas. See Geospan Corp. v. Pictometry Int'l Corp., No. 08-816 ADM/JSM, 2011 WL 1261583, at *3–4 (D. Minn. Mar. 31, 2011) (interpreting the term “based on” in a patent to permit a mathematical combination of the values upon which an image is based).

The Court defined the claim limitation “moving the gradient neighborhood like a sliding window over the 3D elevation data” as “sequentially and completely applying the gradient neighborhood to the 3D elevation.” Claim Construction Order 21. Finally, the Court defined the limitation “is configured to run an algorithm” as “is programmed to run an algorithm without the need to rebuild, rewrite or recompile the code for, or redesign any of that hardware or software.” Id. at 26.

Thus, summary judgment on the issue of infringement of the ’293 Patent is appropriate only if there is no genuine dispute of material fact that the LRAIL system does not implement (i) 3D maps that either merge or blend or separately represent elevation and intensity data, (ii) an algorithm that sequentially and completely applies a sliding gradient window to a 3D map, and (iii) an algorithm that is programmed to run without the need to rebuild, rewrite or recompile the code, or redesign the hardware or software.

C. The Fake3D Image

The parties dispute the properties of the Fake3D Image produced by Plaintiff's LRAIL system. Defendant claims the Fake3D Image is an image containing both elevation and intensity data. Morellas Decl. ¶ 21. Morellas identifies two portions of the source code he claims use elevation data to compute changes in elevation in the x and y axes. Id. ¶ 25. The source code then generates a 3D elevation vector that is normal (orthogonal) to the surface generated by the elevation data. Id. ¶ 26. The source code then computes a dot product ² between the normal elevation vector and two light source vectors. Id. ¶ 27. The source code then multiplies this elevation-light quantity with the intensity data, which results in a representation of elevation, intensity, and light data. Id. ¶ 28. The source code also generates a variable that stores the dot product combination of the three-dimensional elevation and intensity data. Id. ¶ 29.

Plaintiff claims the Fake3D Image contains no elevation data. Hébert Decl. ¶ 35. Instead, Plaintiff argues that the Fake3D Images are 2D intensity images that are enhanced with artificial light sources derived from elevation data. Id. ¶¶ 36–37. Finally, Plaintiff claims the output of the Fake3D Image does not contain the range data. Id. ¶ 37. Hébert argues the source code verifies Plaintiff's position, while ironically not identifying any specific provision of the source code. Id. ¶ 41; see also id. ¶ 1 (“I am responsible for, and actively involved in, directing the development of Pavemetrics’ LRAIL and other products, including their software source code.”).

Morellas also examined the algorithms that analyze the Fake3D Image to identify track features. Morellas Decl. ¶¶ 53–83. He concluded that the LRAIL system identifies features from the Fake3D Image by inputting the Fake3D Image into a neural network. Id. ¶ 59. He also identifies source code that uses a gradient window that analyzes the entire Fake3D Image to identify features. Id. ¶ 69.

Hébert does not dispute that a neural network identifies features from the Fake3D Image. He only concludes that the Fake3D Image is not a “track elevation map” or a “three dimensional elevation map” under his understanding of the claim limitation. See Hébert Decl. ¶ 48. Hébert disputes that the neural network slides a gradient window over the Fake3D Image sequentially and completely because it processes only 50 percent of the image. See Suppl. Hébert Decl. ¶ 31. As a result, Plaintiff argues that the LRAIL system is not “configured” to meet the other claim limitations because meeting the claim limitations would require code modification. See Reply 8.

The Court concludes that Plaintiff fails to demonstrate summary judgment is warranted as to the Fake3D Image. Defendant has put forward evidence tending to show that the Fake3D Image represents elevation and intensity data by blending or merging the individual elevation and intensity values. Morellas Decl. ¶¶ 21–29. Defendant also has put forward evidence demonstrating that the LRAIL system moves a gradient window over the Fake3D Image sequentially and completely. Id. ¶¶ 59, 69. Finally, Defendant has put forward evidence showing the LRAIL system is configured to perform this algorithm without modification. See id. ¶ 69.

Plaintiff disputes the accuracy of this evidence. See Hébert Decl. ¶¶ 35–37, 48; Suppl. Hébert Decl. ¶ 31. The Court cannot grant summary judgment for Plaintiff on this basis. “[I]t is not the province of the district court to weigh conflicting evidence for purposes of summary judgment.” Slenk v. Transworld Sys., Inc., 236 F.3d 1072, 1076 (9th Cir. 2001); accord Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 456, 112 S.Ct. 2072, 119 L.Ed.2d 265 (1992) (noting that the nonmovant's “version of any disputed issue of fact ․ is presumed correct” for summary judgment).

The Court DENIES the motion as to the Fake3D Image.

D. The Fake3D Image with Overlay

Defendant claims the LRAIL system also identifies features from the Fake3D Image by overlaying identifiers over features on the Fake3D Image. Morellas Decl. ¶ 72. Morellas identifies an overlay algorithm that analyzes maps containing only elevation data. Id. ¶¶ 77–80. Morellas bases his conclusion on the fact that the 3D maps identified in claims 1 and 22 only need to contain elevation data. This conclusion is contrary to the Court's definition of the terms “track elevation map” and “three dimensional elevation map.” Claim Construction Order 32. Because Defendant puts forward no evidence creating a material dispute about whether the Fake3D Image with Overlay algorithm identifies features from a 3D map based on both elevation and intensity data, no rational trier of fact could conclude the Fake3D Image with Overlay algorithm meets the “track elevation map” and “three dimensional elevation map” claim limitations. Matsushita, 475 U.S. at 587, 106 S.Ct. 1348.

The Court GRANTS the motion as to the Fake3D Image with Overlay.

E. The Mixed Image

Defendant finally claims the LRAIL system's Mixed Image infringes the ’293 Patent. Morellas claims the Mixed Image contains three data channels containing elevation, intensity, and range data. Morellas Decl. ¶¶ 42–51. Morellas identifies source code that analyzes the Mixed Image without the need for any modification of the code of the LRAIL system. Id. ¶¶ 44, 85.

Plaintiff claims the Mixed Image is a research and development feature developed only for internal testing purposes. Hébert Decl. ¶ 42. Hébert claims Plaintiff “never used a Mixed image to process any data for [its] customers.” Id. He concludes that because no neural networks trained on the Mixed Images exist in the LRAIL system, a neural network identifying features from the Mixed Image would require code modification. Id. ¶ 43. Finally, his supplemental declaration states that no LRAIL systems sent to Plaintiff's customers analyze the Mixed Image. Suppl. Hébert Decl. ¶ 41.

Plaintiff does not adduce evidence disputing that the Mixed Image is a 3D map. Nor does Plaintiff identify any evidence demonstrating that there is no gradient window that is applied sequentially and completely to the Mixed Image. Instead, Plaintiff's only dispute is that the LRAIL system is not configured to identify features from the Mixed Image without code modification. Defendant, however, identifies evidence creating a genuine dispute of material fact as to whether the LRAIL system is configured to identify features from the Mixed Image. Morellas Decl. ¶ 85.

The Court thus DENIES the motion as to the Mixed Image.

IV. CONCLUSION

The Court GRANTS IN PART and DENIES IN PART Plaintiff's motion for summary judgment. The Court determines that the Fake3D Image with Overlay does not infringe independent claims 1 and 22 or dependent claims 15 and 36 of the ’293 Patent. Fed. R. Civ. P. 56(g). The Court denies the motion in all other respects.

The Court provisionally seals this Order. Within seven days of the issuance of this Order, the parties shall file a joint statement as to whether any matter stated in this Order is information that should remain under seal. Thereafter, the Court will determine whether any portions of this Order should be redacted in the version filed on the public docket.

IT IS SO ORDERED.

FOOTNOTES

1.   The Court notes the Supplemental Hébert Declaration is not made under the penalty of perjury. See 28 U.S.C. § 1746.

2.   A dot product of the vectors x = (x1, x2, x3) and y = (y1, y2, y3) is the sum x1 × y1 + x2 × y2 + x3 × y3, where x1, x2, x3, y1, y2, and y3. See Section 5-3: Dot Product, Paul's Online Notes (June 14, 2021), https://tutorial.math.lamar.edu/classes/calcii/dotproduct.aspx.

MARK C. SCARSI, UNITED STATES DISTRICT JUDGE