Training and/or using a recurrent neural network model for visual servoing of an end effector of a robot. In visual servoing, the model can be utilized to generate, at each of a plurality of time steps, an action prediction that represents a prediction of how the end effector should be moved to cause the end effector to move toward a target object. The model can be viewpoint invariant in that it can be utilized across a variety of robots having vision components at a variety of viewpoints and/or can be utilized for a single robot even when a viewpoint, of a vision component of the robot, is drastically altered. Moreover, the model can be trained based on a large quantity of simulated data that is based on simulator(s) performing simulated episode(s) in view of the model. One or more portions of the model can be further trained based on a relatively smaller quantity of real training data.

When a temperature Tedge of a cell at an end of a battery pack is higher than a temperature Tcen of a cell in a pack central portion, a management server generates rebuilding information for rebuilding a battery pack such that a cell less likely to deteriorate than a cell arranged in the pack central portion is arranged at a pack end. When temperature Tcen is higher than temperature Tedge, the management server generates rebuilding information such that a cell less likely to deteriorate than a cell arranged at the pack end is arranged in the pack central portion.

When a temperature Tedge of a cell at an end of a battery pack is higher than a temperature Tcen of a cell in a pack central portion, a management server generates rebuilding information for rebuilding a battery pack such that a cell less likely to deteriorate than a cell arranged in the pack central portion is arranged at a pack end. When temperature Tcen is higher than temperature Tedge, the management server generates rebuilding information such that a cell less likely to deteriorate than a cell arranged at the pack end is arranged in the pack central portion.

A three-dimensional shaping apparatus includes a melting section providing a shaping material to a channel, a nozzle ejecting the shaping material to a shaping region of a shaping table, an ejection-amount adjusting mechanism adjusting an amount of the shaping material from the nozzle, a suction member sucking the shaping material in the channel, a memory configured to store a program, and a processor configured to execute the program so as to control the three-dimensional shaping apparatus. The processor is configured to stop the ejection of the shaping material and, thereafter, prior to resumption of the ejection of the shaping material, execute material purge processing for discharging the shaping material remaining in the melting section to a region different from the shaping region. The suction member is located closer to the nozzle than the ejection-amount adjusting mechanism in the first channel.

A label rewind machine counts labels on a label roll having an elongated web with labels attached. The rewind machine has an unwind spindle and a rewind spindle, and a first label sensor and a second label sensor. The label roll is placed on the unwind spindle, and the web is threaded through the first and second sensors and attached to the rewind spindle, creating a web path. As the web is drawn along the web path, each sensor detects the presence or absence of a label on the web proximate that sensor and generates an electronic signal. Together, the sensors generate a specific quadrature sequence dictated by the presence of labels on the web. If the detected sequence mimics a selected sequence, the rewind machine records a unitary count of one label. The label count is indexed by one each time the rewind machine detects the selected sequence.

A label rewind machine counts labels on a label roll having an elongated web with labels attached. The rewind machine has an unwind spindle and a rewind spindle, and a first label sensor and a second label sensor. The label roll is placed on the unwind spindle, and the web is threaded through the first and second sensors and attached to the rewind spindle, creating a web path. As the web is drawn along the web path, each sensor detects the presence or absence of a label on the web proximate that sensor and generates an electronic signal. Together, the sensors generate a specific quadrature sequence dictated by the presence of labels on the web. If the detected sequence mimics a selected sequence, the rewind machine records a unitary count of one label. The label count is indexed by one each time the rewind machine detects the selected sequence.

A video confirmation computer for confirming video related to robot operation includes a storage unit and a processor. The storage unit stores information on the position of the electric motor that drives a link body of the robot received from the controller of the robot and information on the video obtained by a camera attached to the robot. The processor makes at least one of the video confirmation computer itself and a computer connected to the video confirmation computer display a model area and a video area side by side. In the model area, a two-dimensional or three-dimensional model reproducing the posture of the robot is displayed by computer graphics. In the video area, the video is displayed.

Industrial production facilities (such as can making factories) may tag with radio frequency near proximity devices their tooling (knockouts, dies, punches, etc) so that the tooling may be instantly located on any particular machine or location in the facility, by known critical characteristics of the tooling, particularly diameter. This is possible because each machine will also have a reader for the tags, and other areas such as inventory shelves, inspection stations, doors, loading docks, shipping departments and even vendors may also have the reading devices. Passive RFID tags and readers may be used for this purpose. The invention may be open to other businesses so inventory may be located across manufacturing systems and/or businesses.

A machining apparatus error correction method is implemented in a machining apparatus error correction system. The method includes setting initial operating parameters according to a predetermined machining program, obtaining dimensional detection data during machining of a product, calculating a dimensional correction parameter according to the detection data and a dimensional inspection standard according to a predetermined correction model and generating a correction parameter file readable by the machining apparatus, and distributing the correction parameter file to the corresponding machining apparatus. The initial operating parameters include clamping parameters and dimensional inspection standards.

A machining apparatus error correction method is implemented in a machining apparatus error correction system. The method includes setting initial operating parameters according to a predetermined machining program, obtaining dimensional detection data during machining of a product, calculating a dimensional correction parameter according to the detection data and a dimensional inspection standard according to a predetermined correction model and generating a correction parameter file readable by the machining apparatus, and distributing the correction parameter file to the corresponding machining apparatus. The initial operating parameters include clamping parameters and dimensional inspection standards.