PTAB

IPR2024-00745

Tesla Inc v. Autonomous Devices LLC

Log In
Key Events
Petition

1. Case Identification

2. Patent Overview

  • Title: Artificially Intelligent Systems, Devices, and Methods for Learning and/or Using a Device’s Circumstances for Autonomous Device Operation
  • Brief Description: The ’474 patent discloses techniques for training autonomous devices, such as robots or vehicles. The system allows a "first device" to learn from user-guided operations on a "second device," store this knowledge in a knowledgebase, and then operate autonomously by applying the learned instruction sets when it detects similar circumstances.

3. Grounds for Unpatentability

Ground 1: Obviousness over Buibas and Sinyavskiy - Claims 45-47, 51, 56-57, 59, 68-69 are obvious over Buibas in view of Sinyavskiy.

  • Prior Art Relied Upon: Buibas (Application # 2014/0089232), Sinyavskiy (Application # 2013/0325768).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Buibas discloses a robotic system that meets most limitations of independent claim 45. Buibas teaches using a "robotic brain" or "brain image" (a knowledgebase) which can be a spiking neural network. This brain image receives encoded sensory inputs, processes them, and issues commands (instruction sets) for autonomous operation. Crucially, Buibas teaches that a brain image trained on one robot (a second device) can be shared and loaded onto another robot (a first device), which then applies the learned behaviors. Petitioner asserted that Sinyavskiy remedies a lack of detail in Buibas by explicitly teaching how to partition a spiking neural network for different, dedicated tasks. This partitioning creates a direct correlation between specific inputs and their corresponding instruction sets, as required by the claims.
    • Motivation to Combine: A person of ordinary skill in the art (POSITA) would combine Sinyavskiy with Buibas to improve the functionality of Buibas’s system. While Buibas suggests its neural networks can be partitioned, it provides little detail. Sinyavskiy provides that detail, teaching an advantageous method for partitioning to improve computational efficiency, parallelize task execution, and expand the range of tasks a robot could perform by dedicating partitions to specialized functions.
    • Expectation of Success: A POSITA would have a high expectation of success, as both references describe highly similar systems involving spiking neural networks for training and operating autonomous robots.

Ground 2: Anticipation by Grotmol - Claims 45-47, 51, 59, 68-69 are anticipated by Grotmol.

  • Prior Art Relied Upon: Grotmol (Patent # 9,604,359).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner contended Grotmol discloses every element of the challenged claims. Grotmol describes a robotic system with an "adaptive controller" (knowledgebase) that learns behaviors based on user commands during a training phase. This controller stores correlations between sensory input "features" (a circumstance representation) and motor command "outputs" (instruction sets). Grotmol explicitly teaches that a "trained configuration" learned on one robot (a second device) can be stored and "loaded to one or more other robots" (a first device) to provide those learned behaviors. When the first robot receives the configuration, its controller uses the stored correlations to determine and execute autonomous operations based on its current sensory inputs, thereby meeting all limitations of claim 45.

Ground 3: Obviousness over Grotmol and McClure - Claim 60 is obvious over Grotmol in view of McClure.

  • Prior Art Relied Upon: Grotmol (Patent # 9,604,359), McClure (Patent # 5,219,264).

  • Core Argument for this Ground:

    • Prior Art Mapping: This ground builds on the teachings of Grotmol. Claim 60 adds the limitation that the object representation includes "one or more coordinates" of objects. Petitioner argued that while Grotmol teaches using object "location" to navigate and avoid obstacles, it does not expressly state that this location is represented by coordinates. McClure, however, explicitly discloses a mobile robot that operates using coordinate systems to define the spatial relationship of objects and determines the specific "coordinates" of detected objects in its environment.
    • Motivation to Combine: A POSITA would have been motivated to incorporate McClure’s use of coordinates into Grotmol’s system as a predictable solution to implement Grotmol's teaching of object "location." Using coordinates would increase the robot's "manipulative accuracy" and improve its ability to navigate by allowing for well-known calculations to plot routes and track obstacles, representing a simple application of a known technique to improve a similar system.
    • Expectation of Success: A POSITA would have reasonably expected success, as using coordinate systems to define object locations for robotic navigation was a common and well-known technique at the time.

  • Additional Grounds: Petitioner asserted additional obviousness challenges, including combining Buibas, Sinyavskiy, and Versace to teach the use of object coordinates (Ground 1B), and combining Grotmol with Buibas to teach sharing of trained controllers between devices (Grounds 2C and 2D).

4. Arguments Regarding Discretionary Denial

  • Petitioner argued that discretionary denial under §325(d) is unwarranted because the petition relies on new prior art (Buibas, Sinyavskiy, Versace, Grotmol, McClure) and arguments that were never considered during prosecution.
  • Petitioner further argued that discretionary denial under Fintiv is inappropriate. The co-pending district court litigation involving the ’474 patent has been stayed pending the outcome of this and related IPRs, which strongly weighs in favor of institution. Because of the stay, the Board’s Final Written Decision (FWD) will issue long before any potential trial date, eliminating concerns of inefficiency or duplicative efforts.

5. Relief Requested

  • Petitioner requests institution of an inter partes review (IPR) and cancellation of claims 45-47, 51, 56-57, 59-60, and 68-69 of the ’474 patent as unpatentable.