Systems, devices, and methods for the distribution and collection of multimodal data associated with robots
AssigneeKindred Systems Inc.
InventorsJames Sterling Bergstra, Suzanne Gildert, George Samuel Rose
Life of this patent
Estimating the term…
Abstract
A dynamic representation of a robot in an environment is produced, one or more observer agent collects data, and respective values of one or more metrics for the robot are computed based at least in part on the collected data. Tasks for the robot to perform are generated. Ratings and challenge questions are generated. A server may produce a user interface and a value of a metric based on collected observer data.
Claims20
A method of assessing qualities of a robot, comprising:
- causing, by a processor, a robot to perform a task in an environment;
- capturing, by a camera coupled to the processor, media data of the robot performing the task in the environment, the media data including a video feed;
- generating, by the processor, a dynamic representation of the robot performing the task in the environment from the captured media data;
- providing, by the processor, the dynamic representation of the robot performing the task in the environment to an observer agent;
- requesting, by the processor, a judged level of behavior of the robot, to be input by the observer agent;
- receiving the judged level of behavior of the robot from the observer agent, wherein the judged level of behavior of the robot is determined by the observer agent observing the dynamic representation of the robot performing the task in the environment;
- and
- training a machine learning model using the judged level of behavior of the robot.
A method of assessing qualities of a robot, comprising:
- causing, by a processor, a robot to perform a task in an environment;
- capturing, by a camera coupled to the processor, media data of the robot performing the task in the environment, the media data including a video feed;
- generating, by the processor, a dynamic representation of the robot performing the task in the environment from the captured media data;
- providing, by the processor, the dynamic representation of the robot performing the task in the environment to an observer agent;
- generating, by the processor, a question for the observer agent, wherein the question relates to the appropriateness of the robot's behavior in performing the task;
- receiving, by the processor, an answer to the question from the observer agent, wherein the answer is determined by the observer agent observing the dynamic representation of the robot performing the task in the environment;
- and
- training a machine learning model using the received answer.
A system of assessing qualities of a robot, comprising:
- a robot configured to perform a task in an environment;
- a camera configured to capture media data of the robot performing the task in the environment, the media data including video data;
- a processor communicatively coupled to the robot and the camera, the processor configured to generate a dynamic representation of the robot performing the task in the environment from the captured media data;
- an observer agent communicatively coupled to the processor, the observer agent configured to receive the dynamic representation from the processor;
- and
- an interface coupled to the observer agent, wherein the interface is configured to display a request to input a judged level of behavior of the robot by the observer agent,
- wherein the judged level of behavior of the robot is determined by the observer agent observing the dynamic representation of the robot performing the task in the environment, and wherein the judged level of behavior of the robot is used to train a machine learning model.
Dependencies are inferred from the claim text.