Robotic device and gripping method

1. A robotic device comprising:

an end effector device including two fingers for gripping a workpiece;

a first sensor configured to detect a pressure distribution on a gripping position on the workpiece by the two fingers;

a second sensor configured to detect gripping torque on the workpiece; and

a controller configured to perform, based on a temporal variation in the pressure distribution when the workpiece is lifted, posture control including rotation of the end effector device, wherein

the controller includes a learning section which, based on a plurality of trials including the gripping of the workpiece using the two fingers and movement and rotation of a tool center point of the end effector device, performs supervised learning having been provided as learning data the temporal variation in the pressure distribution, a movement amount, a rotation amount, and a movement speed of the tool center point, and the gripping torque and having been provided as a label a movement amount of the pressure distribution after the rotation, and

the learning section determines, based on a result of the supervised learning, the movement amount and the rotation amount of the tool center point to decrease the movement amount of the pressure distribution after the posture control.

2. The robotic device according to claim 1 ,

wherein

based on a result of the supervised learning having been further provided as learning data the gripping torque prior to the rotation, under a condition where the movement amount of the pressure distribution after the posture control is greater than a first threshold, the learning section determines the gripping torque prior to the posture control to decrease the gripping torque prior to the posture control.

3. The robotic device according to claim 1 , wherein

under a condition where the movement amount of the pressure distribution after the posture control is greater than a second threshold, the learning section determines a movement speed of the tool center point such that the movement speed of the tool center point is high.

4. The robotic device according to claim 1 ,

wherein

based on a result of the supervised learning having been further provided as learning data the gripping torque after the rotation and the temporal variation in the pressure distribution after the rotation, under a condition where the movement amount of the pressure distribution after the posture control is greater than a third threshold, the learning section determines the gripping torque after the posture control to decrease the gripping torque after the posture control.

5. The robotic device according to claim 1 , wherein

when a movement amount of the pressure distribution is greater than a fourth threshold when the workpiece is lifted, the controller performs control so as to retry gripping of the workpiece using the two fingers.

6. A gripping method for gripping a workpiece using a robotic device including an end effector device a first sensor, and a second sensor, the end effector device including two fingers, the first sensor for detecting a pressure distribution on a gripping position on the workpiece by the two fingers, the second sensor for detecting gripping torque on the workpiece, the gripping method comprising:

gripping the workpiece using the two fingers;

detecting a pressure distribution on a gripping position on the workpiece; and

performing, based on a temporal variation in the pressure distribution when the workpiece is lifted, posture control including rotation of the end effector device,

performing, based on a plurality of trials including the gripping of the workpiece using the two fingers and movement and rotation of a tool center point of the end effector device, supervised learning having been provided as learning data the temporal variation in the pressure distribution, a movement amount, a rotation amount, and a movement speed of the tool center point, and the gripping torque and having been provided as a label a movement amount of the pressure distribution after the rotation; and

determining, based on a result of the supervised learning, the movement amount and the rotation amount of the tool center point to decrease the movement amount of the pressure distribution after the posture control.

7. The gripping method according to claim 6 , wherein

the gripping method further comprises determining, based on a result of the supervised learning having been further provided as learning data the gripping torque prior to the rotation, under a condition where the movement amount of the pressure distribution after the posture control is greater than a first threshold, the gripping torque prior to the posture control to decrease the gripping torque prior to the posture control.

8. The gripping method according to claim 6 , further comprising

determining, under a condition where the movement amount of the pressure distribution after the posture control is greater than a second threshold, a movement speed of the tool center point such that the movement speed of the tool center point is high.

9. The gripping method according to claim 6 , wherein

the gripping method further comprises determining, based on a result of the supervised learning having been further provided as learning data the gripping torque after the rotation and the temporal variation in the pressure distribution after the rotation, under a condition where the movement amount of the pressure distribution after the posture control is greater than a third threshold, the gripping torque after the posture control to decrease the gripping torque after the posture control.

10. The gripping method according to claim 6 , further comprising

retrying the gripping of the workpiece using the two fingers when a movement amount of the pressure distribution when the workpiece is lifted is greater than a fourth threshold.