Machine learning device and machine learning method for learning an optimal object gripping path

摘要

A machine learning device for learning an operating state of a robot which stows a plurality of objects arranged on a carrier device in a container using a hand to grasp the objects, the machine learning device comprising: a state observation unit (11) for observing during the operation of the robot , the positions and postures of the objects and a state variable including at least one of a cycle time for stowing the objects in the container and a torque and a vibration occurring when the robot grips the objects; a determination data acquisition unit (12) for obtaining determination data for determining a range of each of the cycle time, the torque and the vibration depending on an allowable tolerance value; anda learning unit (13) for learning the operating state of the robot in accordance with a training data set formed from a combination of the state variable and the determination data.