DCT
2:25-cv-08297
Artificial Intelligence Industry Association Inc v. Elementary Robotics Inc
Key Events
Amended Complaint
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Artificial Intelligence Industry Association, Inc. (Delaware)
- Defendant: Elementary Robotics, Inc. (Delaware)
- Plaintiff’s Counsel: Fry Law Corporation
- Case Identification: 2:25-cv-08297, C.D. Cal., 11/20/2025
- Venue Allegations: Plaintiff alleges venue is proper in the Central District of California because Defendant maintains its principal place of business in South Pasadena, California and has committed alleged acts of infringement in the district by selling accused products to customers with operations there.
- Core Dispute: Plaintiff alleges that Defendant’s AI-powered industrial vision systems infringe three patents related to synthetic image generation for machine learning, stereoscopic 3D video processing, and the embedding of calibration metadata into video files.
- Technical Context: The technology at issue resides in the field of advanced computer vision for industrial automation, where techniques like multi-camera 3D imaging and AI model training with synthetic data are used for high-speed quality inspection.
- Key Procedural History: The complaint alleges that Plaintiff sent a letter to Defendant on June 12, 2025, advising Defendant of its patent portfolio, an event that may be relevant to the allegation of willful infringement.
Case Timeline
| Date | Event |
|---|---|
| 2015-04-29 | Earliest Priority Date for '315 and '693 Patents |
| 2018-09-11 | U.S. Patent No. 10,075,693 Issues |
| 2019-02-12 | U.S. Patent No. 9,930,315 Issues |
| 2019-04-25 | Earliest Priority Date for '272 Patent |
| 2022-03-22 | U.S. Patent No. 11,257,272 Issues |
| 2025-06-12 | Plaintiff sends notice letter to Defendant |
| 2025-11-20 | Complaint Filed |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 9,930,315 - "Recording and Processing of Stereoscopic 3D Video with Embedded Calibration Information"
The Invention Explained
- Problem Addressed: The patent's background describes that prior systems for processing stereoscopic 3D video required separate calibration files to align images from multiple sensors, which created synchronization challenges and computational burdens (Compl. ¶18; ’315 Patent, col. 1:41-50).
- The Patented Solution: The invention proposes a method where calibration information—both static data like lens distortion and time-varying data like motion sensor readings—is embedded directly into the stereoscopic video stream during the recording process. This integrated data is then extracted during playback to perform real-time corrections, such as image stabilization and 3D rendering, without needing separate files (Compl. ¶19; ’315 Patent, col. 3:5-20).
- Technical Importance: This integration of calibration data with the video stream was designed to improve the efficiency and quality of 3D video processing by ensuring that precise alignment and correction data is temporally synchronized with the corresponding video frames (Compl. ¶19).
Key Claims at a Glance
- The complaint asserts independent claim 1 and dependent claim 7 (’315 Patent, Compl. ¶36).
- Claim 1 recites a method with the following essential elements:
- Recording stereoscopic 3D video sequences using a device with two or more image sensors.
- Contemporaneously embedding calibration information (comprising static and time-varying parameters) into the video sequences during recording.
- Storing the video sequences with the embedded information.
- Extracting the embedded information when the video is played back.
- Utilizing the extracted information to perform at least one of correcting distortion, rendering for display, or stabilizing the video sequences.
- The complaint reserves the right to assert additional claims (Compl. ¶36).
U.S. Patent No. 10,075,693 - "Methods and Systems for Embedding Calibration Metadata in Stereoscopic 3D Video Files"
The Invention Explained
- Problem Addressed: The patent addresses technical issues related to the efficient storage and real-time use of calibration data for stereoscopic video, particularly for virtual and augmented reality applications where temporal alignment between frames and metadata is critical (Compl. ¶22, ¶23; ’693 Patent, col. 1:49-2:11).
- The Patented Solution: The invention describes a system comprising a stereoscopic capture device and a processor. The processor is configured to obtain metadata (such as lens parameters and motion sensor data) at the same time as video is recorded, encode this data by embedding it into designated fields within the video file format, and later decode and use this data for tasks like stabilization or projecting the 3D video onto 2D display coordinates (’693 Patent, Abstract; Compl. ¶24).
- Technical Importance: The claimed system aims to reduce computational overhead and enhance the quality of stereoscopic video playback, enabling improved virtual reality experiences (Compl. ¶23).
Key Claims at a Glance
- The complaint asserts independent claims 1 (a system claim) and 11 (a method claim) (Compl. ¶46).
- Claim 1 recites a system with the following essential elements:
- A stereoscopic video capture device.
- A processor configured to: obtain metadata from sensors contemporaneously with recording; encode the metadata into designated metadata fields of the video file format; store the video with the encoded metadata; decode the metadata during playback; and utilize it to perform operations like computing motion vectors, stabilization, or projection transformations.
- The complaint reserves the right to assert additional claims (Compl. ¶46).
U.S. Patent No. 11,257,272 - "Systems and Methods for Generating Synthetic Image Data for Machine Learning Training"
Technology Synopsis
- The patent addresses the costly and time-consuming process of creating large, manually-labeled image datasets needed to train computer vision models. The claimed solution is a system and method for automatically generating realistic, richly annotated synthetic image datasets by constructing virtual 3D scenes, simulating real-world cameras with specific parameters, and using physics-based rendering to create the images and their corresponding ground-truth annotations (Compl. ¶26-27).
Asserted Claims
- Independent claims 1 (a method claim) and 17 (a system claim) are asserted (Compl. ¶50).
Accused Features
- The complaint alleges that the synthetic data generation and "auto-training" capabilities of the VisionStream Platform infringe the ’272 Patent. This includes its alleged functions for creating "infinite variations from live captures to simulate real-world defects," configuring "virtual camera[s] with complete intrinsic/extrinsic parameters," and using "physics-based rendering" to generate training datasets (Compl. ¶52).
III. The Accused Instrumentality
Product Identification
- The "Accused Products" are identified as Defendant’s VisionStream platform, Vision Machine System, and QualityOS Platform (Compl. ¶1).
Functionality and Market Context
- The Accused Products comprise an integrated hardware and software system designed for AI-powered automated quality inspection in manufacturing environments (Compl. ¶31, ¶33). The system allegedly uses multi-camera setups to capture images of production lines, processes these images in real-time on an edge computing device, and uses machine learning models to detect defects (Compl. ¶33).
- Key technical features alleged in the complaint include support for "multi-sensor processing with calibration," the ability to embed calibration and sensor metadata, and systems for "synthetic data generation and augmentation" to expand training datasets for its AI models (Compl. ¶14, ¶33). Defendant is alleged to market these capabilities as enabling "autonomous AI-powered inspection" that achieves high accuracy at high speeds for major industrial clients (Compl. ¶14).
No probative visual evidence provided in complaint.
IV. Analysis of Infringement Allegations
'315 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| recording, by a stereoscopic video capture device comprising two or more image sensors, stereoscopic 3D video sequences | The VisionStream Platform's "multi-sensor processing" capabilities use multiple industrial cameras connected via "dual PoE camera support" to simultaneously capture synchronized video sequences from different viewpoints, creating stereoscopic sequences (Compl. ¶37). | ¶38 | col. 10:30-44 |
| contemporaneously embedding calibration information... into the stereoscopic 3D video sequences during the recording... | The platform is alleged to perform "real-time calibration embedding" of parameters that are "processed on edge," embedding this information into the captured video data streams during recording. | ¶39 | col. 11:5-18 |
| wherein the calibration information comprises static calibration parameters and time-varying calibration parameters | The embedded information allegedly includes static parameters ("complete intrinsic/extrinsic parameters" such as lens distortion) and time-varying parameters ("motion from 1,000 ppm feeds"). | ¶39 | col. 11:7-13 |
| storing the stereoscopic 3D video sequences with the embedded calibration information | The system uses "onboard storage in the Vision Machine hardware" to store captured image sequences with associated metadata, which is deliverable via an API as "sequences with embedded metadata." | ¶40 | col. 11:21-25 |
| extracting... and utilizing the extracted calibration information to perform at least one of: correcting distortion..., rendering..., or stabilizing... | Defendant's platform allegedly performs "motion filtering and stabilization processing... using embedded sensor data," uses "lens distortion" parameters to correct optical distortions, and performs "rendering operations" for display and analysis. | ¶41 | col. 11:26-30 |
- Identified Points of Contention:
- Scope Questions: A central question may be whether the term "stereoscopic 3D video sequences," as used in the patent, can be construed to cover the "multi-perspective imaging" from industrial cameras used for automated quality inspection (Compl. ¶37). The analysis may focus on whether the patent's context limits the term to applications creating depth perception for human viewing, versus the machine vision application alleged here.
- Technical Questions: The complaint alleges "contemporaneously embedding... during the recording" (Compl. ¶39). A factual dispute may arise over the precise timing and mechanism of this embedding. The analysis will likely scrutinize whether the accused system embeds data into the video stream in real-time as it is captured, or whether it associates the data in a separate, post-capture step on the edge device, and whether such a process meets the claim limitation.
'693 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| A system for processing stereoscopic 3D video with embedded metadata, the system comprising: a stereoscopic video capture device... a processor operatively coupled... | The VisionStream Platform is described as an integrated hardware and software system that records stereoscopic video using multiple image sensors and processes the data (Compl. ¶33, ¶47). | ¶47 | col. 3:25-30 |
| obtain metadata from sensors... contemporaneously with recording... wherein the metadata comprises at least one of: lens parameters... intrinsic camera parameters, or motion sensor data | The platform allegedly obtains metadata including "lens parameters," "intrinsic camera parameters," and "motion sensor data" at the same time as recording video sequences (Compl. ¶47). | ¶47 | col. 8:20-31 |
| encode the metadata into a video file format by embedding the metadata into designated metadata fields of the video file format | Defendant is alleged to perform "metadata-embedded video processing" by encoding the obtained metadata into the video file format (Compl. ¶21, ¶47). | ¶47 | col. 8:1-11 |
| store the stereoscopic 3D video sequences with the encoded metadata | The system is alleged to store the video sequences with the encoded metadata (Compl. ¶47). | ¶47 | col. 8:40-42 |
| decode... and utilize the decoded metadata to perform at least one of: computing motion vectors... applying stabilization processing, or performing projection transformations | The complaint alleges the platform decodes the metadata and utilizes it to perform "motion vector computation," "stabilization processing," and "matrix operations for 3D-to-2D projection" (Compl. ¶21, ¶47). | ¶47 | col. 8:2-7 |
- Identified Points of Contention:
- Scope Questions: The term "video file format" may become a focal point. The infringement analysis will question whether the Accused Product's method of providing "API delivery of sequences with embedded metadata" (Compl. ¶17, ¶40) constitutes encoding data into "designated metadata fields" of a "video file format," or if it uses a proprietary data structure that is merely associated with, but not embedded within, a standard video file in the manner contemplated by the patent.
- Technical Questions: Evidence will be required to determine if the accused system's processor performs all the recited functions. For example, while the complaint alleges "3D-to-2D projection" (Compl. ¶21), the defense may argue its system operates differently, raising a factual question about the specific algorithms and transformations being performed.
V. Key Claim Terms for Construction
The Term: "contemporaneously embedding... during the recording" (’315 Patent, Claim 1)
- Context and Importance: This term is critical because the timing of the embedding process is a core element of the invention. Practitioners may focus on this term because the dispute could turn on whether Defendant's "processed on edge" system (Compl. ¶39) performs the embedding in a manner that is functionally simultaneous with the act of capturing video frames, or in a subsequent processing step that may fall outside a narrow construction of the claim.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification's description of a system where calibration can be done in "real time" (e.g., ’315 Patent, col. 5:53-56) may support an interpretation where "contemporaneously" allows for very short, near-real-time delays inherent in a processing pipeline on an edge device.
- Evidence for a Narrower Interpretation: The patent's abstract states that the method embeds information "into the stereoscopic video sequence in a real time as the sequences of stereoscopic images are being recorded." (’315 Patent, Abstract). This could be argued to require that the embedding process be integrated directly with the image capture and encoding hardware, occurring before the data is first stored.
The Term: "designated metadata fields of the video file format" (’693 Patent, Claim 1)
- Context and Importance: This term is central to whether Defendant's method of associating metadata with video data infringes. The case may depend on whether Defendant's data structure, delivered via an API (Compl. ¶17), qualifies as being embedded in "designated fields" of a "format."
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification mentions encoding metadata into "subtitle metadata or a table in the metadata header" (’693 Patent, col. 8:9-11). This suggests flexibility and could support a broader reading that includes various structured data containers associated with the video file, not just rigidly defined standard fields.
- Evidence for a Narrower Interpretation: Claim 11 of the patent refers to a "metadata channel in a video file format" that "comprises a subtitle metadata field or a closed-caption metadata field" (’693 Patent, col. 12:7-12). A party could argue this language, present elsewhere in the patent, implies that "designated metadata fields" refers to specific, pre-defined channels within standard video container formats, not proprietary data structures.
VI. Other Allegations
- Indirect Infringement: The complaint alleges inducement of infringement under 35 U.S.C. § 271(b) based on Defendant's "detailed technical documentation, user manuals, training materials, implementation guides, [and] application notes" that allegedly instruct customers on how to configure and operate the Accused Products in an infringing manner (Compl. ¶56). Contributory infringement is also alleged, based on the assertion that the Accused Products are specially designed for practicing the inventions and have no substantial non-infringing uses (Compl. ¶58).
- Willful Infringement: The complaint alleges willful infringement based on both general and specific knowledge. It asserts that Defendant, as an active participant in the computer vision industry, would have monitored relevant patents (Compl. ¶59). More directly, it alleges Defendant had actual knowledge from a notice letter sent by Plaintiff on or about June 12, 2025, and that Defendant's continued activities after receiving this letter demonstrate willful disregard for Plaintiff's patent rights (Compl. ¶3, ¶60).
VII. Analyst’s Conclusion: Key Questions for the Case
- A core issue will be one of definitional scope: can terms like "stereoscopic 3D video sequences" and "virtual reality experiences," which originate in the context of human visual perception in the asserted patents, be construed to cover the automated, machine-vision processes for industrial quality control performed by the accused system?
- A second issue will be one of claim construction and technical evidence: does the Defendant's method of associating video and calibration data via an "API delivery of sequences with embedded metadata" meet the specific claim requirement of "embedding the metadata into designated metadata fields of the video file format," and does its "auto-training" feature practice the detailed, multi-step synthetic image generation process claimed in the '272 patent?
- A final question will concern willfulness and damages: given the allegation that Defendant received a specific notice letter from Plaintiff months before the suit was filed, the court will have to determine whether any infringement, if found, was willful, which could expose Defendant to the possibility of enhanced damages.