Computing device, machine learning method, and storage medium

A machine learning method for a machine learning model configured such that a computer performs a function of outputting a movement-target object to be moved to a predetermined position at current time in an operation of holding a plurality of target objects with a manipulator of a mobile manipulator and moving the target objects to predetermined positions, the machine learning method comprising:

• training the machine learning model by inputting a plurality of training data sets, which are combinations of state variables and movement-target objects to be moved to predetermined positions at current time, the state variables each containing relative positions of the target objects to a specific portion of the mobile manipulator, the movement-target objects being represented by pieces of determination data and associated with the state variables;
• and
• controlling operation of the mobile manipulator based on the output movement-target object, wherein
• the state variable contains a relative position of an obstacle to the specific portion of the mobile manipulator,
• the relative positions of the target objects to the specific portion of the mobile manipulator and the relative position of the obstacle to the specific portion of the mobile manipulator are represented by data on a map or on a bitmap image, and
• the computer is configured to output data representing a relative position of the movement-target object to the specific portion of the mobile manipulator on the map or on the bitmap image.

A computing device configured to perform computation for controlling operations of a mobile manipulator configured to hold a plurality of target objects with a manipulator and move the target objects to predetermined positions, the computing device comprising:

• a storage configured to store a trained machine learning model trained by inputting a plurality of training data sets, which are combinations of state variables and pieces of determination data associated with the state variables, the training data sets being acquired in advance;
• and
• a calculator configured to output a movement-target object to be moved to a predetermined position at current time by inputting a state variable to the trained machine learning model read from the storage, and
• a controller configured to control operation of the mobile manipulator based on the output movement-target object, wherein
• the state variable contains relative positions of the target objects to a specific portion of the mobile manipulator,
• the determination data associated with the state variable represents the movement-target object,
• the state variable contains route lengths of routes from the mobile manipulator to the target objects, the route lengths being obtained based on a route plan, and
• the state variable contains route lengths of routes from the target objects to predetermined positions associated with the target objects.