A robot control method, and associated robot controllers and robots operating with such methods and controllers, providing real-time vibration suppression. The control method involves learning to support real-time, vibration-suppressing control. The method uses state-of-the-art machine learning techniques in conjunction with a differentiable dynamics simulator to yield fast and accurate vibration suppression. Vibration suppression using offline simulation approaches that can be computationally expensive may be used to create training data for the controller, which may be provide by a variety of neural network configurations. In other cases, sensory feedback from sensors onboard the robot being controlled can be used to provide training data to account for wear of the robot's components.

When workpiece (W) is brought into a non-gripping state after deflection compensation of robot arm (10) is performed in a gripping state of workpiece (W), the deflection compensation of robot arm (10) is performed in a non-gripping state of workpiece (W). Here, the deflection compensation of robot arm (10) in the non-gripping state of workpiece (W) is performed while a compensation amount is changed to gradually decrease, while hand (18) is moved from a first teaching point to a second teaching point.

The present disclosure generally relates to the field of robotics and computer animation, more particularly, method and apparatus to solve the inverse kinematics problem to control a kinematic chain such as a robot arm or an animation character's skeleton to reach a target position. The new method simulates a kinematic chain whose links and joints are elastic and can be distorted. The method distorts the kinematic chain to move its end to the target position, calculates distortions, and iteratively adjusts link and joint configurations of the kinematic chain to reduce distortions while keeping its end at the target position until a solution with near zero distortions is found. The resulting link and joint configurations of the simulated kinematic chain then can be used for the actual kinematic chain to reach the same target position.

A method for characterising the performance of a joint in a surgical robotic arm, the joint being driven by a drivetrain which transfers power from a drive source to the joint, the method comprising: sending a first command signal to position the robot arm into an initial configuration; sending a second command signal to apply a force to the joint to displace the joint from a steady state; for a plurality of predefined time intervals: receiving a first measurement indicating the configuration of the drive source at a first end of the drivetrain; receiving a second measurement indicating the configuration of the joint at a second end of the drivetrain; calculating a value of elongation using the first and second measurements; and receiving a third measurement indicating the torque experienced by the joint at the second end of the drivetrain; comparing the values of elongation with corresponding values of torque at each of the predefined time intervals; and generating an output from the comparison indicating the performance of the joint.

The present disclosure generally relates to the field of robotics and computer animation, more particularly, method and apparatus to solve the inverse kinematics problem to control a kinematic chain such as a robot arm or an animation character's skeleton to reach a target position. The new method simulates a kinematic chain whose links and joints are elastic and can be distorted. The method distorts the kinematic chain to move its end to the target position, calculates distortions, and iteratively adjusts link and joint configurations of the kinematic chain to reduce distortions while keeping its end at the target position until a solution with near zero distortions is found. The resulting link and joint configurations of the simulated kinematic chain then can be used for the actual kinematic chain to reach the same target position.

Provided is a robot control device capable of reducing a robot vibration amount using machine learning based on a small number of operations. A robot control device according to one aspect of the present invention that, in order to perform a task in relation to a target object which is made to move by a robot, controls operation by the robot based on an operation program that uses a plurality of pass-through points to specify a movement path that includes one or more task sections in which the task is to be performed, the robot control device including: a command value generation unit configured to, based on the operation program, generates a command value that instructs a state of the robot for each time; a driving unit configured to drive the robot in accordance with the command value; a vibration amount obtainment unit configured to, for each time, obtain an amount of vibration of the robot that is driven by the driving unit; a vibration amount extraction unit configured to, based on the operation program, extract the amount of vibration for a time corresponding to the task section from among the amounts of vibration obtained by the vibration amount obtainment unit; and a command value correction unit configured to, based on the amount of vibration extracted by the vibration amount extraction unit, correct the command value.

A robotic arm having one or more hybrid (soft-rigid) joints includes a first link, a second link, and a joint interconnecting the first link and the second link, such that the first link is movable relative to the second link along an axis of motion. The joint includes: a socket component coupled to a distal end portion of the second link and a ball component coupled to a proximal end portion of first link, the ball component is configured to rotationally fit within the socket component. The joint also includes a flexible membrane encasing the socket component and the ball component. The robotic arm is controllable using a reinforcement learning algorithm training using a simulation of the robotic arm and optionally, further training in a physical world.

A method of obtaining the gear stiffness of a robot joint gear of a robot joint of a robot arm, where the robot joint is connectable to at least another robot joint. The robot joint comprises a joint motor having a motor axle configured to rotate an output axle via the robot joint gear. The method comprises the steps of: —applying a motor torque to the motor axle using the joint motor; —obtaining the angular position of the motor axle; —obtaining the angular position of the output axle; —determining the gear stiffness based on at least the angular position of the motor axle, the angular position of the output axle and a dynamic model of the robot arm.

A method of controlling a robot having a plurality of joints includes measuring load torque applied to a driving-force transmission system of each of the plurality of joints while moving a hand of the robot along a predetermined path, comparing a measurement value of the load torque and an allowable range of each of the joints, and controlling a rate of change in acceleration of the driving-force transmission system of each of the joints, depending on a comparison result, in a next operation in which the hand of the robot is moved along the predetermined path.

Motion control of a robot arm is performed via a reducer connected to a motor. A controller thereof includes a thrust control unit that generates motor position command value based on an input thrust command value, and a motor control unit that generates a current value based on the motor position command value. The motor control unit feeds back a motor position detected by a motor encoder, and the thrust control unit feeds back thrust detected by a thrust meter. The feedback from the motor control unit suppresses vibration phenomena at the reducer, and the feedback from the thrust control unit suppresses transmission error, thereby enabling motion control of the arm with rapidity and precision.