US 11126889

Here's a concise summary of US Patent 11126889:

US Patent 11126889B2

• Title: Machine learning based prediction of human interactions with autonomous vehicles
• Assignee: Perceptive Automata LLC (Current Assignee as of 2025-02-19). The original assignee was Perceptive Automata Inc.
• Inventors: Samuel English Anthony, Kshitij Misra, Avery Wagner Faller
• Filing Date: March 24, 2020
• Issue Date: September 21, 2021 (Application granted on this date).
• Abstract: Systems and methods for predicting user interaction with vehicles are disclosed. A computing device receives image or video segments of a road scene from a participant's perspective, then generates stimulus data including original or altered versions of these segments. This stimulus data is transmitted to a user interface to collect response data, such as an action or its likelihood, from users regarding other road participants. The computing device aggregates this response data into statistical data to create a model. This model is then applied to new image or video segments to generate predictions of user behavior in those segments.

Plain-Language Overview of Independent Claims:

The patent contains two independent claims: Claim 1 and Claim 9.

• Claim 1 (Method for predicting user interaction with vehicles): This claim describes a method involving a computing device that performs several steps:

  1. Receive Visual Data: The device gets images or video of a road scene (from a vehicle's perspective) that include pedestrians, cyclists, or other vehicles.
  2. Create Stimulus Data: It then generates special "stimulus data" which can be the original visual data or a modified version of it.
  3. Send to User Interface: This stimulus data is sent to a display for users.
  4. Collect Human Responses: The device receives feedback from users (human observers) indicating what they think the people/vehicles in the scene will do, or how likely they are to do it. This feedback is gathered via the user interface.
  5. Aggregate Responses: The collected human responses for specific images/videos are combined to create statistical data.
  6. Develop Prediction Model: A machine learning model is built using this statistical data.
  7. Apply Model for Prediction: This trained model is then used on new images or videos (e.g., live road scenes) to predict the behavior of road users in those new scenes.

• Claim 9 (System for predicting user interaction with vehicles): This claim describes a system that performs the same functions as the method in Claim 1, but frames it as a computing device with specific components (e.g., one or more processors and memory) configured to execute those steps:

  1. Receive Visual Data Module: A component for getting images or video of a road scene.
  2. Generate Stimulus Data Module: A component for creating original or altered stimulus data from the visual input.
  3. Transmit Stimulus Data Module: A component for sending stimulus data to a user interface.
  4. Receive Response Data Module: A component for collecting human feedback (actions and likelihoods) from the user interface.
  5. Aggregate Response Data Module: A component for compiling statistical data from the human responses.
  6. Create and Apply Model Module: A component for building a machine learning model from the statistical data and using it to predict user behavior in new scenes.

CAFC 2026 Dockets:

As of April 26, 2026, a search of CAFC 2026 dockets did not return any cases specifically mentioning patent number US11126889. The patent's status information indicates ongoing litigation including a PTAB case (IPR2025-01574) and US District Court cases in Texas (7:25-cv-00594 and 2:25-cv-00742), but no direct CAFC dockets for 2026 were found during the search.