PTAB

IPR2023-01173

Tesla Inc v. Autonomous Devices LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: System for Autonomous Device Operation via Visual Learning
  • Brief Description: The ’583 patent discloses systems and methods for device automation. The core technology involves a system learning correlations between digital pictures of a device's surroundings and corresponding instruction sets for operating the device, storing this learned information in a knowledgebase, and using it to operate autonomously.

3. Grounds for Unpatentability

Ground 1: Obviousness of Claims 10 and 11 over Grotmol and POSITA Knowledge

  • Prior Art Relied Upon: Grotmol (Patent 9,604,359) and the general knowledge of a Person of Ordinary Skill in the Art (POSITA).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Grotmol discloses the core limitations of the challenged claims. Grotmol teaches a robotic system that, during a training phase, learns associations between captured images of its environment and motor commands issued by a trainer. This data (training sets) is stored in a memory buffer. For autonomous operation, the robot captures a new image, compares it to the stored images to find the best match, and then executes the associated motor command. This process maps to the claimed system of learning picture-instruction correlations and using them for autonomous operation. For claim 10, which requires learning from two different users, Petitioner asserted this would be an obvious extension, as Grotmol teaches training both simple and complex tasks, making it logical for a POSITA to use trainers with varying skills. For claim 11, which requires storing correlations for operating different devices, Petitioner pointed to Grotmol’s disclosure that a "trained configuration may be loaded to one or more other robots," making this an obvious implementation.
    • Motivation to Combine: Not applicable (single reference ground). The motivation was framed as obvious design choices for a POSITA to improve training efficiency and versatility.
    • Expectation of Success: A POSITA would have reasonably expected success in using multiple trainers or sharing training data between robots, as these are straightforward methods to enhance the robustness and scope of a learning-based system like Grotmol's.

Ground 2: Obviousness of Claims 25 and 27 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: Grotmol provided the foundational autonomous system. Petitioner argued that Hickman supplies the missing elements for claims 25 and 27, which relate to a shared, server-based knowledgebase. Hickman discloses "shared robot knowledge bases for use with cloud computing systems" where information learned by one robot is uploaded to a central server. This collective knowledge can then be accessed by other robots, allowing them to "benefit from the prior experiences of other robots." This directly teaches the server-based architecture of claim 25, where a first device sends data to a server and a second device receives it. It also teaches the multi-device knowledgebase of claim 27, where the knowledgebase includes correlations learned from operating different devices.
    • Motivation to Combine: A POSITA would combine Grotmol's autonomous learning system with Hickman's shared cloud knowledgebase to achieve significant advantages. Grotmol already teaches sharing trained configurations between robots. Hickman provides a known, scalable, and powerful method (a cloud platform) to implement and enhance this sharing. The combination would allow robots to learn collectively, improving performance on complex tasks (like "fetch," which is discussed in both patents) and leveraging the processing power of the cloud.
    • Expectation of Success: The combination was argued to be predictable. Hickman's cloud architecture is designed specifically to aggregate and distribute the type of experiential data generated by a learning robot like the one disclosed in Grotmol.

Ground 3: Obviousness of Claim 30 over Grotmol in view of Hoffman

  • Prior Art Relied Upon: Grotmol (Patent 9,604,359) and Hoffman (Patent 9,283,674).
  • Core Argument for this Ground:
    • Prior Art Mapping: Grotmol provided the base system, including its methods for object recognition within captured images. Claim 30 adds the limitation that the system generates a "top-down view" of object representations. Petitioner contended that Grotmol does not explicitly disclose a top-down view. However, Hoffman was argued to cure this deficiency by teaching a method of operating a robot using a remote controller that displays an augmented map of the robot's surroundings. This map is explicitly described and depicted as a top-down view showing the location of the robot and obstacles.
    • Motivation to Combine: A POSITA would have been motivated to integrate Hoffman’s display technology into Grotmol’s system. Grotmol already discloses using a remote device (e.g., a smartphone) for controlling and training its robot. Hoffman teaches an improved user interface for such a remote device, providing a top-down map for superior situational awareness. A POSITA would have recognized this as a known technique to improve a similar device (Grotmol's controller) in a predictable way, providing more detailed information to the user.
    • Expectation of Success: Integrating a known visualization method (a top-down map) into an existing remote-control interface would have been a straightforward enhancement with a high expectation of success.

4. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 10, 11, 25, 27, and 30 of the ’583 patent as unpatentable.