Simulation of liquid objects manipulated by robots using supervised learning

A computer-readable storage medium storing a simulation program causing a processor included in a simulation device to:

• set framing conditions for a model representing an operation subject, wherein the operation subject is liquid in a container, and wherein the framing conditions are physical conditions for moving the model representing the operation subject in a simulation space in a same manner as in reality and comprise conditions representing flexibility or viscosity of the operation subject;
• set conditions for external force applied to the operation subject comprising conditions for force added by a robot and force added by a boundary of the container;
• simulate movement of the operation subject under the framing conditions and the conditions for external force;
• generate learning data including the framing conditions, the conditions for external force, and data representing movement of a plurality of representative points located on a liquid surface at a plurality of positions in a perpendicular direction during simulation of the movement of the operation subject;
• by supervised learning using the learning data, generate a learning model that takes the framing conditions, the conditions for external force, and initial conditions of the plurality of representative points located on the liquid surface as input and outputs data representing movement of the plurality of representative points located on the liquid surface;
• simulate a movement of a robot;
• input the framing conditions and conditions for external force added to the operation subject by the robot during simulation of the movement of the robot to the learning model and predict movement of the plurality of representative points located on the liquid surface;
• and
• combine the predicted movement of the plurality of representative points located on the liquid surface and the simulated movement of the operation subject.

A simulation device comprising:

• a processor, configured to:
• set framing conditions for a model representing an operation subject, wherein the operation subject is liquid in a container, and wherein the framing conditions are physical conditions for moving the model representing the operation subject in a simulation space in a same manner as in reality and comprise conditions representing flexibility or viscosity of the operation subject;
• set conditions for external force applied to the operation subject comprising conditions for force added by a robot and force added by a boundary of the container;
• simulate movement of the operation subject under the framing conditions and the conditions for external force;
• generate learning data including the framing conditions, the conditions for external force, and data representing movement of a plurality of representative points located on a liquid surface at a plurality of positions in a perpendicular direction during simulation of the movement of the operation subject;
• by supervised learning using the learning data, generate a learning model that takes the framing conditions, the conditions for external force, and initial conditions of the plurality of representative points located on the liquid surface as input and outputs data representing movement of the plurality of representative points located on the liquid surface;
• simulate a movement of a robot;
• input the framing conditions and conditions for external force added to the operation subject by the robot during simulation of the movement of the robot to the learning model and predict movement of the plurality of representative points located on the liquid surface;
• and
• combine the predicted movement of the plurality of representative points located on the liquid surface and the simulated movement of the operation subject.