Device and method for training a neural network for controlling a robot for an inserting task

A method for inserting an object into an insertion, comprising:

• controlling the robot to hold the object;
• controlling the robot to move in a direction of a plane in which the insertion is located until the object or a part of the object touches the plane;
• controlling the robot to press the object onto the plane;
• measuring the force and the moment experienced by the object;
• feeding the measured force and measured moment to a neural network trained by:
• for each first position of a plurality of first positions in which the object or the part of the object held by the robot touches a plane in which the insertion is located:
• controlling the robot to move to the first position,
• controlling the robot to press the object onto the plane,
• measuring the force and the moment experienced by the object,
• scaling a pair formed of the force and moment by a number randomly chosen between zero and a predetermined positive maximum number, and
• labelling the scaled pair of force and moment by a movement vector between the position and the insertion;
• and
• training the neural network using the labelled pairs of force and moment;
• and
• controlling the robot to move according to a movement vector derived by the neural network from the camera image.

A robot, comprising:

• a camera mounted on the robot adapted to provide camera images;
• and
• a controller configured to implement a neural network and configured to train a neural network to derive, from a force and a moment exerted on an object when pressed on a plane in which an insertion for inserting the object is located, a movement vector to insert an object into an insertion, wherein for training the neural network the controller is configured to:
• for each position of a plurality of positions in which the object or a part of the object held by the robot touches a plane in which the insertion is located:
• control the robot to move to a position of the plurality of positions,
• control the robot to press the object onto the plane,
• measure the force and the moment experienced by the object,
• scale a pair formed of the measured force and moment by a number randomly chosen between zero and a predetermined positive maximum number, and
• label the scaled pair of force and moment by a movement vector between the position and the insertion;
• and
• train the neural network using the labelled pairs of force and moment.

A non-transitory computer readable medium on which are stored storing instructions for training a neural network to derive, from a force and a moment exerted on an object when pressed on a plane in which an insertion for inserting the object is located, a movement vector to insert an object into an insertion, the instructions, when executed by a processor, causing the processor to perform the following steps:

• for each position of a plurality of positions in which the object or a part of the object held by the robot touches a plane in which the insertion is located:
• controlling the robot to move to a position of the plurality of positions,
• controlling the robot to press the object onto the plane,
• measuring the force and the moment experienced by the object,
• scaling a pair formed of the measured force and moment by a number randomly chosen between zero and a predetermined positive maximum number, and
• labelling the scaled pair of force and moment by a movement vector between the position and the insertion;
• and
• training the neural network using the labelled pairs of force and moment.