Reinforcement learning using obfuscated environment models

A system comprising:

• one or more computers;
• and
• one or more storage devices communicatively coupled to the one or more computers, wherein the one or more storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations for training an action selection system used to select actions to be performed by an agent interacting with a target environment to perform a task in the target environment, the operations comprising:
• obtaining, by the one or more computers, a target environment model of the target environment;
• modifying, by the one or more computers, the target environment model of the target environment to generate an obfuscated environment model of an obfuscated environment that represents an obfuscation of the target environment;
• receiving, by the one or more computers and from each of a plurality of users, a request to interact with the obfuscated environment through an obfuscated environment simulation that implements the obfuscated environment model;
• providing, by the one or more computers and to each of the plurality of users, access to a corresponding obfuscated environment simulation that implements the obfuscated environment model;
• obtaining, by the one or more computers and from each of the plurality of users, one or more obfuscated environment trajectories that represent interaction of the user with the obfuscated environment through the corresponding obfuscated environment simulation;
• mapping, by the one or more computers, each of the obfuscated environment trajectories representing interaction with the obfuscated environment to a corresponding target environment trajectory representing interaction with the target environment;
• and
• training, by the one or more computers, the action selection system used to select actions to be performed by the agent interacting with the target environment to accomplish the task in the target environment on the target environment trajectories.

One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations for training an action selection system used to select actions to be performed by an agent interacting with a target environment to perform a task in the target environment, the operations comprising:

• obtaining, by the one or more computers, a target environment model of the target environment;
• modifying, by the one or more computers, the target environment model of the target environment to generate an obfuscated environment model of an obfuscated environment that represents an obfuscation of the target environment;
• receiving, by the one or more computers and from each of a plurality of users, a request to interact with the obfuscated environment through an obfuscated environment simulation that implements the obfuscated environment model;
• providing, by the one or more computers and to each of the plurality of users, access to a corresponding obfuscated environment simulation that implements the obfuscated environment model;
• obtaining, by the one or more computers and from each of the plurality of users, one or more obfuscated environment trajectories that represent interaction of the user with the obfuscated environment through the corresponding obfuscated environment simulation;
• mapping, by the one or more computers, each of the obfuscated environment trajectories representing interaction with the obfuscated environment to a corresponding target environment trajectory representing interaction with the target environment;
• and
• training, by the one or more computers, the action selection system used to select actions to be performed by the agent interacting with the target environment to accomplish the task in the target environment on the target environment trajectories.