Tesla, Inc. v. Autonomous Devices, LLC

1. Case Identification

• Patent #: 10,452,974
• Filed: June 9, 2023
• Petitioner(s): Tesla, Inc.
• Patent Owner(s): Autonomous Devices, LLC
• Challenged Claims: 1-20

2. Patent Overview

• Title: Autonomous Device Operation via Learned Circumstances
• Brief Description: The ’974 patent relates to systems and methods for autonomous device operation. The core technology involves a device learning from a user's operation by creating and storing correlations between "circumstance representations" (e.g., images or sensor data) and corresponding "instruction sets" in a knowledgebase. The system then autonomously operates by anticipating and executing these learned instructions when it encounters new, partially matching circumstances.

3. Grounds for Unpatentability

Ground 1: Anticipation over Grotmol - Claims 1-9 and 12-20 are anticipated by Grotmol.

• Prior Art Relied Upon: Grotmol (Patent 9,604,359).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Grotmol discloses all limitations of the challenged claims. Grotmol teaches a robotic system that learns associations between "sensory context" (images, which are the claimed "circumstance representations") and a "respective action" (motor commands, which are the claimed "instruction sets") during a user-operated training phase. This data is stored in a memory buffer (the claimed "knowledgebase"). During autonomous operation, the system compares a current, newly acquired image (the "third circumstance representation") to the stored images. Upon finding a "best matching stored image" (a "partial match"), it retrieves and executes the associated stored motor commands. Grotmol also explicitly discloses that a robot's "trained configuration" can be stored and "loaded to one or more other robots," thereby anticipating limitations related to a "second device."

Ground 2: Obviousness over Grotmol and POSITA Knowledge - Claims 10 and 11 are obvious over Grotmol in view of the knowledge of a POSITA.

• Prior Art Relied Upon: Grotmol (Patent 9,604,359).
• Core Argument for this Ground:
  • Prior Art Mapping: Claims 10 and 11 add the requirement that a correlation is learned in an additional learning process involving "another user" operating the first device (claim 10) or a second device (claim 11). Petitioner contended that Grotmol discloses training a robot for multiple behaviors, including both "simple" and "complex tasks," through multiple learning processes.
  • Motivation to Combine (N/A - POSITA Knowledge): A person of ordinary skill in the art (POSITA) would have found it obvious to use different users (trainers) for these multiple training processes to take advantage of the varying skills and knowledge of different trainers. Grotmol does not limit the number of users who can train the robot. Therefore, Petitioner argued it would have been obvious for an additional learning process to be performed by "another user" on either the first device or a second device, which Grotmol teaches can be loaded with the initial training.

Ground 3: Obviousness over Grotmol and Hickman - Claims 4 and 11 are obvious over Grotmol in view of Hickman.

• Prior Art Relied Upon: Grotmol (Patent 9,604,359) and Hickman (Patent 8,639,644).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Grotmol teaches the fundamental system of a robot learning from a user and storing correlations of circumstances and instructions. Grotmol further discloses that this learned training can be shared with other robots. Hickman discloses a "shared robot knowledge base" for use with cloud computing systems, where learned experiences from one robot are uploaded to a central repository to benefit other robots. The combination of Grotmol's learning robot with Hickman's cloud-based shared knowledgebase would result in a system where correlations are learned from the operation of multiple devices (e.g., a "fourth correlation" from a "second device" as required by claim 4, and training by another user on a second device as per claim 11).
  • Motivation to Combine: A POSITA would combine the two references to improve the system. Grotmol expressly contemplated sharing trained configurations among robots, and Hickman teaches a known and advantageous method—a cloud-based shared knowledgebase—to facilitate this sharing on a large scale. The combination would predictably enhance Grotmol's system with Hickman's superior networking and data-sharing capabilities. Petitioner noted that Hickman is cited on the face of Grotmol and that the assignee of Grotmol publicly discussed the benefits of such shared robot learning.
  • Expectation of Success: A POSITA would have a reasonable expectation of success in combining the references. Hickman's system is designed to store and distribute the exact type of data—learned robot behaviors based on experiences—that Grotmol's system generates, making the integration straightforward and the outcome predictable.

4. Key Claim Construction Positions

• "means for generating or receiving," "means for anticipating," and "means for executing" (Claim 19): Petitioner argued that these means-plus-function limitations require construction under 35 U.S.C. §112(f). Based on the '974 patent's specification, which states that "one or more processor circuits" perform these functions, Petitioner contended that the corresponding structure for all three "means for" limitations is "one or more processor circuits." This construction is central to Petitioner's argument that the processor-based systems in the prior art meet these limitations.

5. Relief Requested

• Petitioner requests institution of an inter partes review and cancellation of claims 1-20 of the '974 patent as unpatentable.