A method creates numerical control data sets for controlling machine tools. The control data sets are read from the machine tools. A first component data set representing a first component design model is received. A first numerical control data set is created for the first component data set using control program generation software, having an assessment routine using a trained machine learning algorithm with settable parameters. A first additional training data set is compiled from the component data set and the created numerical control data set. The first additional training data set is output to a training database. The machine learning algorithm is updated by setting usage-environment-specific values for the parameters determined by training the machine learning training algorithm using the training database.

During machining of a workpiece, a gripping force adjustment device takes into account the state of the machining and the state of the workpiece in order to set a more appropriate gripping force. The gripping force adjustment device acquires data indicating a machining state implemented by a machine tool and data relating to a gripping state realized on the workpiece by a jig, and creates data to be used in machine learning on the basis of the acquired data. The gripping force adjustment device then executes machine learning processing relating to the gripping force exerted on the workpiece by the jig in the environment in which the machine tool machines the workpiece on the basis of the created data.

Disclosed is a reconfigurable device for dispensing/distributing tablets in the blister pockets of a packaging strip subjected to longitudinal movement in a substantially horizontal plane of a packaging machine having a blister-packer, the reconfigurable device having a frame designed to be mounted on the blister-packer, the frame having a front attachment wall preferably arranged vertically, at least one accessory for dispensing/distributing the tablets in the blister pockets of the strip, the accessory being designed to be mounted removably on the frame. The device includes an intermediate interface intended to be attached removably to the frame, against the front attachment wall thereof, in order to support the at least one accessory. The interface includes at least one drive member of a mobile element of the accessory/accessories supported thereby.

A machine learning device for performing machine learning with respect to a numerical control device which causes a machine tool to operate, and is provided with: a state information acquisition unit that causes the machine tool to perform cutting work, in which a cutting amount and a cutting rate are set, and acquires state information including the cutting amount and cutting rate; an action information output unit that outputs action information; a reward calculation unit that acquires determination information that is information about the strength of pressure applied to a tool at least during cutting work, the shape of the waveform of the pressure applied to the tool, and time it has taken to perform work, and outputs a reward value in reinforcement learning; and a value function update unit that updates a value function on the basis of the reward value, the state information, and the action information.

A tendency of an action of a robot may vary based on learning data used for training. The learning data may be generated by an agent performing an identical or similar task to a task of the robot. An apparatus and method for updating a policy for controlling an action of a robot may update the policy of the robot using a plurality of learning data sets generated by a plurality of heterogeneous agents, such that the robot may appropriately act even in an unpredicted environment.

A robot manipulator including limbs moveable via bearings controlled by actuators; sensors to capture a bearing position and a bearing torque/bearing force; a first sensor to capture a force screw W; a housing downstream of the first sensor; a second sensor to capture a user force applied to the housing and/or a user torque; a computing unit to determine, using a dynamics model of the robot manipulator and based on particular bearing torque/bearing force, the force screw W, and the user force and/or the user torque, a first force and/or a first torque to shift the limbs and a second force and/or a second torque to apply to an external object via an effector, wherein the dynamics model includes at least gravitational forces and inertial forces based on the bearing position; and a storage unit to store the first and/or the second force, and/or the first and/or the second torque.

A method for executing a process, in particular using at least one robot, includes executing a run-through of the process, detecting a value of a first process variable, and detecting an assessment of this executed process run-through. Assessment learning steps are then repeated multiple times, wherein run-throughs of the process using varied process controls are executed and additional assessments are detected. A first quality factor model of the process, which model determines a quality factor for the process on the basis on the first process variable, is machine-learned based on the detected assessments and values of the first process variable. The method further includes repeating process control optimization steps multiple times.

A Bayesian learning approach for stability boundary and optimal parameter identification in milling without the knowledge of the underlying tool dynamics or material cutting force coefficients. Different axial depth and spindle speed combinations are characterized by a probability of stability which is updated based upon whether the result is stable or unstable. A likelihood function incorporates knowledge of stability behavior. Numerical results show convergence to an analytical stability lobe diagram. An adaptive experimental strategy identifies optimal operating parameters that maximize material removal rate. An efficient and robust learning method to identify the stability lobe diagram and optimal operating parameters with a limited number of tests/data points.

A set condition determining method for manufacturing facilities includes: inputting, into a trained model, a manufacturing condition for a target product and a setup condition that is for a product manufactured in same manufacturing facilities before manufacture of the target product and that reflects setup condition modification by an operator's manual manipulation; and obtaining a setup condition for the target product. The trained model has been trained with input being: manufacturing conditions for the target product; and setup conditions that are for the product manufactured in the same manufacturing facilities before the manufacture of the target product and that reflect setup condition modification by an operator's manual manipulation, and output being setup conditions for the target product.

A state determination device that determines an operation state of an injection molding machine stores respective specification data of a reference injection molding machine and an injection molding machine that is different from the reference injection molding machine, and acquires data related to the injection molding machine. Then, the state determination device converts the acquired data into yardstick data by a conversion formula set for every type of data, by using the stored specification data of the reference injection molding machine and the stored specification data of the injection molding machine and performs machine learning using the yardstick data obtained through the conversion so as to generate a learning model.