Robot control device, robot system and robot control method

A robot control method for controlling a robot based on a result of predicting output data from input data using machine learning, the input data being a state of the robot and its surroundings when a human operates the robot so that the robot performs a series of operations, the output data being the corresponding human operation or an operation of the robot by the human operation, the robot control method comprising:

• a provisional operation information output step for outputting provisional operation information for operating the robot provisionally to make the robot perform the series of operations;
• a provisional control step for making the robot perform the series of operations by controlling the robot based on the provisional operation information, and for causing a human to operate the robot to intervene in the provisional operation for modifying the provisional operation;
• a modification work data collection step for collecting modification work data which is to be trained by a training model, the modification work data including input data and output data, the input data being the state of the robot and its surroundings when the human operates the robot to intervene in the provisional operation by the provisional control step so as to modify the provisional operation, the output data being data indicating the corresponding human operation for modifying the provisional operation or a modification operation of the robot by the human operation for modifying the provisional operation;
• a trained model building step for building a trained model by training the modification work data to the training model;
• and
• a modified control step for controlling the robot based on the provisional operation information and for applying the output data output by the trained model to a control of the robot when the state of the robot and its surroundings is input to the trained model as the input data.