Amazon Technologies, Inc.

Corrinne Yu

Brian C Beckman

Brandon William Porter

Michael Vogelsong

Benjamin Lev Snyder

Leonardo Ruggiero Bachega

Gur Kimchi

John Clark Coonley Duksta

A machine learning system builds and uses computer models for controlling robotic performance of a task. Such computer models may be first trained using feedback on computer simulations of the robotic system performing the task, and then refined using feedback on real-world trials of the robot performing the task. Some examples of the computer models can be trained to automatically evaluate robotic task performance and provide the feedback. This feedback can be used by a machine learning system, for example an evolution strategies system or reinforcement learning system, to generate and refine the controller.

Simulation-real world feedback loop for learning robotic control policies

Disclosed are systems and methods for reducing the amount of messaging between aerial vehicles and between controllers of aerial vehicles and simplifying aerial vehicle traffic management. In one implementation, a large service area, such as the United States, may be separated into a series of hierarchal regions. Rather than sending notifications to all agents (e.g., aerial vehicles, controllers) in the service area, each agent may subscribe to one or more regions of the hierarchal regions and only receive messages intended for the subscribed regions. In one example, as discussed below, messages for a particular region are only sent to agents subscribed to that region. Other agents within the larger service area do not receive the messages as they may not be relevant to those agents.

Vehicle path hash values

A machine learning system builds and uses computer models for controlling robotic performance of a task. Such computer models may be first trained using feedback on computer simulations of the robot performing the task, and then refined using feedback on real-world trials of the robot performing the task. Some examples of the computer models can be trained to automatically evaluate robotic task performance and provide the feedback. This feedback can be used by a machine learning system, for example an evolution strategies system or reinforcement learning system, to generate and refine the controller.

Artificial intelligence system for learning robotic control policies

A machine learning system builds and uses computer models for identifying how to evaluate the level of success reflected in a recorded observation of a task. Such computer models may be used to generate a policy for controlling a robotic system performing the task. The computer models can also be used to evaluate robotic task performance and provide feedback for recalibrating the robotic control policy.

Artificial intelligence system for modeling and evaluating robotic success at task performance

Hierarchal reference system

Reducing status notifications using a hierarchal reference system

Disclosed are systems and methods for storing path data for all vehicles of a service area in a path data store and representing each path with a unique hash value, generated based on the path data. Rather than controllers and/or aerial vehicles exchanging the full path data for each path, a common hash function may be used to generate unique hash values for each path and the unique hash values may be exchanged and used to lookup the full path data.

Growing community of inventors

Seattle, WA, United States of America

Corrinne Yu