A method of training a robot system for manipulation of objects, the robot system being able to perform a set of skills, wherein each skill is learned as a skill model, the method comprising: receiving physical input from a human trainer, regarding the skill to be learned by the robot; determining for the skill model a set of task parameters including determining for each task parameter of the set of task parameters if a task parameter is an attached task parameter, which is related to an object being part of said kinesthetic demonstration or if a task parameter is a free task parameter, which is not related to a physical object; obtaining data for each task parameter of the set of task parameters from the set of kinesthetic demonstrations, and training the skill model with the set of task parameters and the data obtained for each task parameter.

A cross laser calibration device used to calibrate a tool center point is provided. The calibration device includes a coordinate orifice plate, a set of cross laser sensors and a rotational and translational movement mechanism. The coordinate orifice plate has an orifice center point. The set of cross laser sensors is arranged on the coordinate orifice plate to generate cross laser lines intersecting at the orifice center point. The set of cross laser sensors is driven by the second motor to rotate around the center point of the second motor, wherein the orifice center point has an off-axis setting relative to the center point of the second motor.

A calibration device is provided. The calibration device includes a frame, a first optical sensing device, a second optical sensing device and a third optical sensing device. The frame includes a bottom plate and at least four sidewalls, wherein the sidewalls have a first grating hole, a second grating hole, a third grating hole and a fourth grating hole at a first height. The bottom plate has an image recognition pattern, a first measurement point, a second measurement point and a third measurement point.

A calibration system for robot tool including a robot arm adopting a first coordinate system, a tool arranged on a flange of the robot arm, and an imaging device adopting a second coordinate system is disclosed, wherein an image sensing area is established by the image device. A calibration method is also disclosed and includes steps of: controlling the robot arm to move for leading a tool working point (TWP) of the tool enters the image sensing area; recording a current gesture of the robot arm as well as a specific coordinate of the TWP currently in the second coordinate system; obtaining a transformation matrix previously established for describing a relationship between the first and the second coordinate systems; and importing the specific coordinate and the current gesture to the transformation matrix for calculating an absolute position of the TWP in the first coordinate system.

A calibration apparatus includes a processor, an alignment device, and an arm. The alignment device captures images in a three-dimensional space, and a tool is arranged on a flange of the arm. The processor records a first matrix of transformation between an end-effector coordinate-system and a robot coordinate-system, and performs a tool calibration procedure according to the images captured by the alignment device for obtaining a second matrix of transformation between a tool coordinate-system and the end-effector coordinate-system. The processor calculates relative position of a tool center point of the tool in the robot coordinate-system based on the first and second matrixes, and controls the TCP to move in the three-dimensional space for performing a positioning procedure so as to regard points in an alignment device coordinate-system as points of the TCP, and calculates the relative positions of points in the alignment device coordinate-system and in the robot coordinate-system.

A method of training a robot system for manipulation of objects, the robot system being able to perform a set of skills, wherein each skill is learned as a skill model, the method comprising: receiving physical input from a human trainer, regarding the skill to be learned by the robot; determining for the skill model a set of task parameters including determining for each task parameter of the set of task parameters if a task parameter is an attached task parameter, which is related to an object being part of said kinesthetic demonstration or if a task parameter is a free task parameter, which is not related to a physical object; obtaining data for each task parameter of the set of task parameters from the set of kinesthetic demonstrations, and training the skill model with the set of task parameters and the data obtained for each task parameter.

A position correction method includes: a step of opposing a hand to a target by moving the hand such that the hand becomes in a predetermined first initial posture; a first position detection step of detecting a position of the target from a rotation angle of a rotation axis when the target blocks a detection light by swinging the hand; a step of opposing the hand to the target by moving the hand such that the hand becomes a predetermined second initial posture different from the first initial posture; a second position detection step of detecting a position of the target from the rotation angle of the rotation axis when the target blocks the detection light by swinging the hand; and a correction amount arithmetic step of obtaining rotation angle correction amounts of the second axis and the third axis based on a difference between the position of the target acquired in the first initial posture and the position of the target acquired in the second initial posture.

A calibration device is provided. The calibration device includes a frame, a first optical sensing device, a second optical sensing device and a third optical sensing device. The frame includes a bottom plate and at least four sidewalls, wherein the sidewalls have a first grating hole, a second grating hole, a third grating hole and a fourth grating hole at a first height. The bottom plate has an image recognition pattern, a first measurement point, a second measurement point and a third measurement point.

A position correction method includes: a step of opposing a hand to a target by moving the hand such that the hand becomes in a predetermined first initial posture; a first position detection step of detecting a position of the target from a rotation angle of a rotation axis when the target blocks a detection light by swinging the hand; a step of opposing the hand to the target by moving the hand such that the hand becomes a predetermined second initial posture different from the first initial posture; a second position detection step of detecting a position of the target from the rotation angle of the rotation axis when the target blocks the detection light by swinging the hand; and a correction amount arithmetic step of obtaining rotation angle correction amounts of the second axis and the third axis based on a difference between the position of the target acquired in the first initial posture and the position of the target acquired in the second initial posture.

A calibration system for robot tool including a robot arm adopting a first coordinate system, a tool arranged on a flange of the robot arm, and an imaging device adopting a second coordinate system is disclosed, wherein an image sensing area is established by the image device. A calibration method is also disclosed and includes steps of: controlling the robot arm to move for leading a tool working point (TWP) of the tool enters the image sensing area; recording a current gesture of the robot arm as well as a specific coordinate of the TWP currently in the second coordinate system; obtaining a transformation matrix previously established for describing a relationship between the first and the second coordinate systems; and importing the specific coordinate and the current gesture to the transformation matrix for calculating an absolute position of the TWP in the first coordinate system.