公开(公告)号：US11780083B2

公开(公告)日：2023-10-10

申请号：US17520175

申请日：2021-11-05

申请人： GOOGLE LLC

发明人： Nicolas Hudson ,  Devesh Yamparala

IPC分类号： B25J13/08 ,  B25J9/16 ,  G05B13/02 ,  G06N3/084 ,  G06N3/008 ,  G06N3/045

CPC分类号： B25J9/161 ,  B25J9/1602 ,  B25J9/163 ,  B25J9/1656 ,  B25J9/1697 ,  G05B13/027 ,  G06N3/008 ,  G06N3/045 ,  G06N3/084 ,  G05B2219/33036 ,  G05B2219/33037

摘要： Methods, apparatus, and computer-readable media for determining and utilizing human corrections to robot actions. In some implementations, in response to determining a human correction of a robot action, a correction instance is generated that includes sensor data, captured by one or more sensors of the robot, that is relevant to the corrected action. The correction instance can further include determined incorrect parameter(s) utilized in performing the robot action and/or correction information that is based on the human correction. The correction instance can be utilized to generate training example(s) for training one or model(s), such as neural network model(s), corresponding to those used in determining the incorrect parameter(s). In various implementations, the training is based on correction instances from multiple robots. After a revised version of a model is generated, the revised version can thereafter be utilized by one or more of the multiple robots.

公开(公告)号：US10054930B2

公开(公告)日：2018-08-21

申请号：US14491715

申请日：2014-09-19

申请人： FANUC Corporation

发明人： Zhaojia Liu

IPC分类号： G05B19/404

CPC分类号： G05B19/404 ,  G05B2219/33036 ,  G05B2219/49344

摘要： In an error correction amount creating device that creates an error correction amount for a five-axis machine controlled by a numerical controller and having three linear axes and two rotation axes, the translation error correction amount and the rotation error correction amount for each lattice point of a lattice region into which a two-dimensional coordinate system space with the rotation axes is divided is obtained from data measured for each position of division of the respective axes and given to the numerical controller.

公开(公告)号：US20100094788A1

公开(公告)日：2010-04-15

申请号：US12522040

申请日：2007-12-19

申请人： Anton Maximilian Schäfer ,  Steffen Udluft ,  Hans-Georg Zimmerman

发明人： Anton Maximilian Schäfer ,  Steffen Udluft ,  Hans-Georg Zimmerman

IPC分类号： G06N3/08 ,  G05B13/04

CPC分类号： G05B13/027 ,  G05B2219/33011 ,  G05B2219/33021 ,  G05B2219/33025 ,  G05B2219/33036 ,  G06N3/02

摘要： A method for the computer-assisted control and/or regulation of a technical system is provided. The method includes two steps, namely learning the dynamics of a technical system using historical data based on a recurrent neuronal network, and the subsequent learning of an optimum regulation by coupling the recurrent neuronal network to another neuronal network. The method can be used with any technical system in order to control the system in an optimum computer-assisted manner. For example, the method can be used in the control of a gas turbine.

摘要翻译： 提供了一种用于计算机辅助控制和/或调节技术系统的方法。 该方法包括两个步骤，即使用基于复发性神经元网络的历史数据来学习技术系统的动力学，以及随后通过将复发性神经元网络耦合到另一神经元网络来学习最佳调节。 该方法可以与任何技术系统一起使用，以便以最佳的计算机辅助方式控制系统。 例如，该方法可用于燃气轮机的控制。

公开(公告)号：US5548192A

公开(公告)日：1996-08-20

申请号：US490478

申请日：1995-06-14

申请人： D. Mitchel Hanks

发明人： D. Mitchel Hanks

IPC分类号： G05B13/02 ,  G05B19/19 ,  G05D3/12 ,  G05B19/00

CPC分类号： G05B13/027 ,  G05B19/19 ,  G05B2219/33036 ,  G05B2219/33046 ,  G05B2219/41151 ,  G05B2219/41152

摘要： A method for adjusting a compensator in a feedback control system for a coupled plant, derives a plurality of forward error values from position error signals that are sensed as outputs from the plant in response to a disturbance of the plant's operation. The position error signals are determined over plural, increasingly forward, time segments. A plurality of compensator output values are also determined over predetermined plural increasing forward, time segments. The derived forward error values are applied as inputs to the plant in a back propagation order, that order being in reverse time order to the time order in which the forward position error values were derived. Plural back propagation error values result in response to the back propagation application of the forward position error values. The back propagation error values are derived from within a portion of the compensator which has been configured as a Direct Form II filter function. The plural back propagation error values are then utilized in conjunction with the compensator output values and the forward position error values to derive altered compensator gain stage values to enable a reduction of the forward error values. To provide increased correction capability, the forward position error signals are altered by weighting functions before being employed as inputs to achieve the back propagation error values.

摘要翻译： 一种用于调整用于耦合设备的反馈控制系统中的补偿器的方法，从响应于设备操作干扰的来自工厂的位置误差信号导出多个前向误差值。 位置误差信号在多个，日益前进的时间段上确定。 还在预定的多个增加的前进时间段中确定多个补偿器输出值。 导出的前向误差值以反向传播顺序作为输入施加到工厂，该顺序是与向前位置误差值的时间顺序相反的时间顺序。 多个反向传播误差值导致响应于前向位置误差值的反向传播应用。 反向传播误差值是从已经被配置为直接形式II滤波器功能的补偿器的一部分内导出的。 然后，多个反向传播误差值与补偿器输出值和前向位置误差值结合使用以导出改变的补偿器增益级值，以便能够减少前向误差值。 为了提供增加的校正能力，在被用作输入以实现反向传播误差值之前，通过加权函数来改变前向位置误差信号。

公开(公告)号：US20190001489A1

公开(公告)日：2019-01-03

申请号：US15640914

申请日：2017-07-03

申请人： X Development LLC

发明人： Nicolas Hudson ,  Devesh Yamparala

IPC分类号： B25J9/16 ,  G05B13/02

CPC分类号： B25J9/161 ,  B25J9/1602 ,  B25J9/163 ,  B25J9/1656 ,  B25J9/1697 ,  G05B13/027 ,  G05B2219/33036 ,  G05B2219/33037 ,  G06N3/008 ,  G06N3/0454 ,  G06N3/084

摘要： Methods, apparatus, and computer-readable media for determining and utilizing human corrections to robot actions. In some implementations, in response to determining a human correction of a robot action, a correction instance is generated that includes sensor data, captured by one or more sensors of the robot, that is relevant to the corrected action. The correction instance can further include determined incorrect parameter(s) utilized in performing the robot action and/or correction information that is based on the human correction. The correction instance can be utilized to generate training example(s) for training one or model(s), such as neural network model(s), corresponding to those used in determining the incorrect parameter(s). In various implementations, the training is based on correction instances from multiple robots. After a revised version of a model is generated, the revised version can thereafter be utilized by one or more of the multiple robots.

公开(公告)号：US20100049339A1

公开(公告)日：2010-02-25

申请号：US12521920

申请日：2007-12-19

申请人： Anton Maximilian Schäfer ,  Steffen Udluft

发明人： Anton Maximilian Schäfer ,  Steffen Udluft

IPC分类号： G05B13/04 ,  G06N3/08

CPC分类号： G05B13/027 ,  G05B2219/33025 ,  G05B2219/33036 ,  G06N3/08

摘要： A method for the computer-assisted control and/or regulation of a technical system is provided. The method is used to efficiently reduce a high-dimensional state space describing the technical system to a smaller dimension. The reduction of the state space is performed using an artificial recurrent neuronal network. In addition, the reduction of the state space enables conventional learning methods, which are only designed for small dimensions of state spaces, to be applied to complex technical systems with an initially large state space, wherein the conventional learning methods are performed in the reduced state space. The method can be used with any technical system, especially gas turbines.

摘要翻译： 提供了一种用于计算机辅助控制和/或调节技术系统的方法。 该方法用于将描述技术系统的高维状态空间有效地降低到更小的维度。 使用人工复发神经元网络来执行状态空间的减少。 此外，状态空间的减少使得仅能够设计用于状态空间的小尺寸的常规学习方法被应用于具有最初大的状态空间的复杂技术系统，其中传统的学习方法以简化的状态执行 空间。 该方法可用于任何技术系统，特别是燃气轮机。

公开(公告)号：US08566264B2

公开(公告)日：2013-10-22

申请号：US12521920

申请日：2007-12-19

申请人： Anton Maximilian Schäfer ,  Steffen Udluft

发明人： Anton Maximilian Schäfer ,  Steffen Udluft

IPC分类号： G06E1/00 ,  G06E3/00 ,  G06F15/18 ,  G06G7/00 ,  G06N3/02

CPC分类号： G05B13/027 ,  G05B2219/33025 ,  G05B2219/33036 ,  G06N3/08

摘要： A method for the computer-assisted control and/or regulation of a technical system is provided. The method is used to efficiently reduce a high-dimensional state space describing the technical system to a smaller dimension. The reduction of the state space is performed using an artificial recurrent neuronal network. In addition, the reduction of the state space enables conventional learning methods, which are only designed for small dimensions of state spaces, to be applied to complex technical systems with an initially large state space, wherein the conventional learning methods are performed in the reduced state space. The method can be used with any technical system, especially gas turbines.

摘要翻译： 提供了一种用于计算机辅助控制和/或调节技术系统的方法。 该方法用于将描述技术系统的高维状态空间有效地降低到更小的维度。 使用人工复发神经元网络来执行状态空间的减少。 此外，状态空间的减少使得仅能够设计用于状态空间的小尺寸的常规学习方法被应用于具有最初大的状态空间的复杂技术系统，其中传统的学习方法以简化的状态执行 空间。 该方法可用于任何技术系统，特别是燃气轮机。

公开(公告)号：US06560498B1

公开(公告)日：2003-05-06

申请号：US09623228

申请日：2000-08-30

申请人： Jong Gye Shin ,  Won Don Kim

发明人： Jong Gye Shin ,  Won Don Kim

IPC分类号： G06F1900

CPC分类号： G05B19/4099 ,  G05B2219/33018 ,  G05B2219/33036 ,  G05B2219/33037 ,  G05B2219/36284 ,  G05B2219/45143

摘要： This invention includes the generation of forming information and its manipulation scheme as a method to form curved plates in ship hull-pieces. This invention consists of three components as follows: one is to construct and utilize a database which includes data about flat plates, objective curved plates, plates which are being formed, and their forming information, another is to infer new forming information with an artificial neural network system, and the third is to obtain forming information through calculating in-plane and bending strains. In the third, initial forming information is obtained by calculating strains from relationship between flat plates and objective curved plates. And new forming information is yielded through calculating the strains from relationship between partially formed curved plates and objective curved plates. Final objective plate are reached by repeatedly performing the measurement of the difference between plates in the proceeding steps and final objective plates and the calculation of the new strains in each process. Therefore, through this invention standardization and automation can be realized in the formation of curved plates.

摘要翻译： 本发明包括形成信息的生成及其操作方案，作为在船体中形成弯曲板的方法。 本发明由以下三个组成部分组成：一个是构建和利用数据库，该数据库包括关于平板，目标曲面板，正在形成的板的数据及其形成信息，另一个是用人造神经 网络系统，第三是通过计算平面和弯曲应变来获得形成信息。 第三，通过从平板和客观曲面板之间的关系计算应变来获得初始成形信息。 并且通过从部分形成的弯曲板和客观弯曲板之间的关系计算应变来产生新的成形信息。 通过反复进行步骤和最终目标板之间的差异的测量以及每个过程中新应变的计算来达到最终目标板。 因此，通过本发明，可以在曲面板的形成中实现标准化和自动化。

公开(公告)号：US5579442A

公开(公告)日：1996-11-26

申请号：US233423

申请日：1994-04-28

申请人： Takashi Kimoto ,  Daiki Masumoto ,  Hiroshi Yamakawa ,  Shigemi Nagata

发明人： Takashi Kimoto ,  Daiki Masumoto ,  Hiroshi Yamakawa ,  Shigemi Nagata

IPC分类号： G05B13/02 ,  B25J9/16 ,  B25J9/18 ,  G06F15/18 ,  G06N99/00 ,  G05B19/04

CPC分类号： B25J9/161 ,  B25J9/1697 ,  G05B2219/33036 ,  G05B2219/39268 ,  G05B2219/39295 ,  G05B2219/40564 ,  G05B2219/40617

摘要： A network is provided in which the direct model of the controlled target is learned, and an actuation amount for realizing a target control amount is generated using this network in accordance with the relaxation algorithm. A recognizing network and a prediction network are provided, wherein the recognition result of the recognizing network is input to the prediction network, and the prediction result of the predicting network is fed back to the recognizing network. Moreover, the predicting network is utilized so as to generate a command relating to the motion in accordance with the relaxation algorithm.

摘要翻译： 提供一种网络，其中学习受控对象的直接模型，并且根据松弛算法使用该网络生成用于实现目标控制量的激活量。 提供识别网络和预测网络，其中将识别网络的识别结果输入到预测网络，并将预测网络的预测结果反馈给识别网络。 此外，使用预测网络，以便根据松弛算法产生与运动相关的命令。

公开(公告)号：US20230405812A1

公开(公告)日：2023-12-21

申请号：US18239735

申请日：2023-08-29

申请人： GOOGLE LLC

发明人： Nicolas Hudson ,  Devesh Yamparala

IPC分类号： B25J9/16 ,  G05B13/02 ,  G06N3/084 ,  G06N3/008 ,  G06N3/045

CPC分类号： B25J9/161 ,  B25J9/163 ,  G05B13/027 ,  B25J9/1656 ,  B25J9/1602 ,  B25J9/1697 ,  G06N3/084 ,  G06N3/008 ,  G06N3/045 ,  G05B2219/33036 ,  G05B2219/33037

摘要： Methods, apparatus, and computer-readable media for determining and utilizing human corrections to robot actions. In some implementations, in response to determining a human correction of a robot action, a correction instance is generated that includes sensor data, captured by one or more sensors of the robot, that is relevant to the corrected action. The correction instance can further include determined incorrect parameter(s) utilized in performing the robot action and/or correction information that is based on the human correction. The correction instance can be utilized to generate training example(s) for training one or model(s), such as neural network model(s), corresponding to those used in determining the incorrect parameter(s).