Robot controller that controls robot, learned model, method of controlling robot, and storage medium

A learned model, comprising:

• one or more memories configured to store computer-executable instructions;
• and
• one or more processors configured to execute the computer-executable instructions stored in the one or more memories to acquire the learned model that is acquired by updating a neural network according to a reward generated in a case where a plurality of virtual images photographed while changing an environmental condition of a virtual environment generated by virtualizing an environment including a robot, which has an arm having multiple joints, and a state of a virtual robot are input to the neural network, and a policy of the virtual robot, which is output from the neural network, satisfies a predetermined condition,
• wherein the policy relates to an action of the arm, the action indicating respective driving amounts of the multiple joints.

A method of controlling a robot, which has an arm having multiple joints, the method comprising:

• acquiring an image from an image capturing apparatus that photographs an environment including the robot;
• driving the robot based on an output result obtained by inputting the image to a neural network;
• and
• updating the neural network according to a reward generated in a case where virtual images photographed while changing an environmental condition of a virtual environment generated by virtualizing the environment and a state of a virtual robot are input to the neural network, and a policy of the virtual robot, which is output from the neural network, satisfies a predetermined condition, and
• wherein the policy relates to an action of the arm, the action indicating respective driving amounts of the multiple joints.

A non-transitory computer-readable storage medium storing a computer-executable program for executing a method of controlling a robot, which has an arm having multiple joints,

• wherein the method comprises:
• acquiring an image from an image capturing apparatus that photographs an environment including the robot;
• driving the robot based on an output result obtained by inputting the image to a neural network;
• and
• updating the neural network according to a reward generated in a case where virtual images photographed while changing an environmental condition of a virtual environment generated by virtualizing the environment and a state of a virtual robot are input to the neural network, and a policy of the virtual robot, which is output from the neural network, satisfies a predetermined condition, and
• wherein the policy relates to an action of the arm, the action indicating respective driving amounts of the multiple joints.