专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明专利

JP2010287028A Information processor, information processing method and program 审中-公开
标题翻译：信息处理器，信息处理方法和程序
公开(公告)号：JP2010287028A
公开(公告)日：2010-12-24
申请号：JP2009140065
申请日：2009-06-11
申请人： Sony Corp , ソニー株式会社
发明人： YOSHIIKE YUKIKO , KAWAMOTO KENTA , NODA KUNIAKI , SABE KOTARO
IPC分类号： G06N5/04
CPC分类号： G06N3/006 , G06K9/00335 , G06K9/00664 , G06K9/6297
摘要： PROBLEM TO BE SOLVED: To determine a proper action of an agent. SOLUTION: In the information processor, a state recognition part 23 calculates a current-state series candidate that is a state series for an agent capable of actions reaching the current state, based on a Hidden Markov Model (HMM) stipulated by a state transition probability that a state will be transitioned according to each of actions performed by an agent capable of actions, and an observation probability that a predetermined observation value will be observed from the state, using an action performed by the agent. An action determination part 24 determines an action to be performed next using the current-state series candidate in accordance with a predetermined strategy.This invention is applicable to an agent that performs autonomous action. COPYRIGHT: (C)2011,JPO&INPIT
摘要翻译：要解决的问题：确定代理人的正确行为。解决方案：在信息处理器中，状态识别部23基于由a（a）所规定的隐马尔可夫模型（HMM），计算作为达到当前状态的动作的代理的状态序列的当前状态序列候选状态转移概率，状态将根据由能够执行动作的代理执行的动作而被转换，以及使用代理执行的动作从状态观察到预定观察值的观察概率。动作确定部24根据预定策略，使用当前状态序列候选来确定下一个要执行的动作。本发明适用于执行自主动作的代理。版权所有（C）2011，JPO＆INPIT

2. 发明专利

JP2007280067A Information processor, processing method and program 审中-公开
标题翻译：信息处理器，处理方法和程序
公开(公告)号：JP2007280067A
公开(公告)日：2007-10-25
申请号：JP2006105702
申请日：2006-04-06
申请人： Sony Corp , ソニー株式会社
发明人： SUZUKI HIROTAKA , MINAMINO KATSUKI , ITO MASATO , YOSHIIKE YUKIKO , KAWAMOTO KENTA
IPC分类号： G06N3/02
摘要： PROBLEM TO BE SOLVED: To suppress discontinuity of signals outputted by recognition generation processing. SOLUTION: A generation part 211 generates an output signal corresponding to time-series data supplied from a characteristic extraction part 13 by use of a node of a dynamics storage network designated by a control signal 18. The generation part 211 smoothes, prior to output of the generated output signal, the signal according to a recognition result of whether the current state is a known state by learning or an unknown state without learning. The information processing method is applicable to, for example, an information processor for recognizing and generating time-series data. COPYRIGHT: (C)2008,JPO&INPIT
摘要翻译：要解决的问题：抑制通过识别生成处理输出的信号的不连续性。解决方案：生成部211通过使用由控制信号18指定的动态存储网络的节点，生成与从特征提取部13提供的时间序列数据对应的输出信号。生成部211平滑化输出所生成的输出信号，该信号根据当前状态是否通过学习已知状态或未学习的未知状态的识别结果。该信息处理方法适用于例如用于识别和生成时间序列数据的信息处理器。版权所有（C）2008，JPO＆INPIT

3. 发明专利

JP2007280066A Information processor, processing method and program 审中-公开
标题翻译：信息处理器，处理方法和程序
公开(公告)号：JP2007280066A
公开(公告)日：2007-10-25
申请号：JP2006105701
申请日：2006-04-06
申请人： Sony Corp , ソニー株式会社
发明人： SUZUKI HIROTAKA , MINAMINO KATSUKI , ITO MASATO , YOSHIIKE YUKIKO , KAWAMOTO KENTA
IPC分类号： G06N3/02
摘要： PROBLEM TO BE SOLVED: To suppress discontinuity of signals outputted by recognition generation processing. SOLUTION: A recognition part 11 determines a winner node corresponding to the dynamics most suitable to time-series data supplied from a characteristic extraction part 13 from nodes of a dynamics storage network, and outputs information showing the determined winner node as a recognition result 17. A recognition part 111 determines the winner node so as to suppress discontinuity of output signals caused by switching of winner node. The information processing method is applicable to, for example, an information processor for recognizing and generating time-series data. COPYRIGHT: (C)2008,JPO&INPIT
摘要翻译：要解决的问题：抑制通过识别生成处理输出的信号的不连续性。解决方案：识别部分11从动态存储网络的节点确定对应于最适合于从特征提取部分13提供的时间序列数据的动力的胜利者节点，并且将表示确定的胜者结点的信息作为识别输出结果17.识别部分111确定胜者节点，以便抑制由优胜者节点的切换引起的输出信号的不连续性。该信息处理方法适用于例如用于识别和生成时间序列数据的信息处理器。版权所有（C）2008，JPO＆INPIT

4. 发明专利

JP2005199402A Behavior control system and method of robot device 有权
标题翻译：行为控制系统和机器人装置的方法
公开(公告)号：JP2005199402A
公开(公告)日：2005-07-28
申请号：JP2004009689
申请日：2004-01-16
申请人： Sony Corp , ソニー株式会社
发明人： NODA KUNIAKI , OTANI NOBUYA , SAWADA TSUTOMU , YOSHIIKE YUKIKO , FUJITA MASAHIRO
IPC分类号： B25J13/00 , B25J5/00
CPC分类号： B25J11/001 , G06N3/008
摘要： PROBLEM TO BE SOLVED: To provide a behavior control system and method of a robot device having a function adaptably changing over between a behavior selecting standard of taking account of one's own state and a behavior selecting standard of taking account of the state of the others according to the situation, which is required for an autonomous robot device. SOLUTION: A behavior selection control system in the robot device is provided with a situated behavior layer (SBL) capable of selecting a specific behavior out of a plurality of behavior and outputting it, and an AL calculation part 120 calculating an activation level (AL) indicating an execution priority of each behavior for selecting the behavior. The AL calculation part 120 is provided with a one's own AL calculation part and an others' ALs calculation part 124 calculating the one's own AL and the others' ALs which indicate the execution priority of the behavior based on the oneself and the others, or interaction objects, ; and an AL integration part 125 weightingly adding the one's own AL and the others' ALs by parameters for determining whether giving greater importance to the one's own state or giving greater importance to the others' states and outputting a final AL. COPYRIGHT: (C)2005,JPO&NCIPI
摘要翻译：要解决的问题：提供一种机器人装置的行为控制系统和方法，其具有在考虑到自己的状态的行为选择标准和考虑到自身状态的行为选择标准之间适应地变化的功能其他根据情况，这是自主机器人设备所需要的。解决方案：机器人装置中的行为选择控制系统设置有能够从多种行为中选择特定行为并输出的位置行为层（SBL），并且AL计算部分120计算激活水平（AL）指示用于选择行为的每个行为的执行优先级。 AL计算部分120具有自己的AL计算部分和其他的ALs计算部分124，其计算自己的AL，以及基于自己和其他人指示行为的执行优先级的其他AL，或者相互作用对象; 和AL集成部分125通过参数来加重自己的AL和其他AL，以确定是否更重视自己的状态或者更加重视他人的状态并输出最终的AL。版权所有（C）2005，JPO＆NCIPI

5. 发明专利

JP2008276290A Data processing apparatus, data processing method, and program 审中-公开
标题翻译：数据处理设备，数据处理方法和程序
公开(公告)号：JP2008276290A
公开(公告)日：2008-11-13
申请号：JP2007115693
申请日：2007-04-25
申请人： Sony Corp , ソニー株式会社
发明人： MINAMINO KATSUKI , AOYAMA KAZUMI , YOSHIIKE YUKIKO , SHIMOMURA HIDEKI
IPC分类号： G06N3/08 , G06N3/00 , G10L15/16
摘要： PROBLEM TO BE SOLVED: To properly learn time-series data additionally in unsupervised learning. SOLUTION: A network management part 7 manages the scale of a time-series pattern storage network composed of a plurality of nodes holding time-series patterns, i.e., patterns of time-series data. A generating part 6 generates the time-series data of the time-series patterns held by the nodes of the time-series pattern storage network. A learning part 4 uses new time-series data, i.e., time-series data to be observed from the outside, and generated time-series data, i.e., time-series data generated by the generating part 6, as time-series data for updating used in the self-organizing updating of the time-series pattern storage network, and updates the time-series pattern storage network in a self-organizing manner using the time-series data for updating. COPYRIGHT: (C)2009,JPO&INPIT
摘要翻译：要解决的问题：另外在无人值守学习中适当地学习时间序列数据。解决方案：网络管理部分7管理由保持时间序列模式的多个节点组成的时间序列模式存储网络的规模，即时间序列数据的模式。生成部6生成由时间序列模式存储网络的节点保持的时间序列模式的时间序列数据。学习部分4使用新的时间序列数据，即从外部观察的时间序列数据，并且生成时间序列数据，即由生成部分6生成的时间序列数据作为时间序列数据，作为时间序列数据，在时间序列模式存储网络的自组织更新中使用的更新，并且使用用于更新的时间序列数据以自组织的方式更新时间序列模式存储网络。版权所有（C）2009，JPO＆INPIT

6. 发明专利

JP2006023952A Robot device and its content management method 有权
标题翻译：机器人及其内容管理方法
公开(公告)号：JP2006023952A
公开(公告)日：2006-01-26
申请号：JP2004201013
申请日：2004-07-07
申请人： Sony Corp , ソニー株式会社
发明人： DI PROFIO UGO , YOSHIIKE YUKIKO
IPC分类号： G06F17/30 , B25J5/00 , B25J13/00 , G06F12/00
摘要： PROBLEM TO BE SOLVED: To provide a robot device which manages acquired content with a more biological method, and also provide its content management method.
SOLUTION: When a content request is generated in a request module 10, an acquisition module 20 searches the Internet to acquire the content, and a filter unit 23 filters the content based on information from a management module 30 and an internal state module 50. A content multidimensional unit 26 converts the filtered content to four levels, that is, a detail level, a summary level, an existence level and a deletion level, to store them in a content storage part 40. A lifetime unit 32 monitors the lifetime of the stored content. A content deletion unit 33 deletes the content of deletion level or less importance and makes the filter unit 23 register the information on the deleted content.
COPYRIGHT: (C)2006,JPO&NCIPI
摘要翻译：要解决的问题：提供一种以更具生物学方法管理所获取的内容的机器人装置，并且还提供其内容管理方法。解决方案：当在请求模块10中产生内容请求时，获取模块20搜索因特网以获得内容，并且过滤器单元23基于来自管理模块30和内部状态模块的信息对内容进行过滤内容多维单元26将过滤的内容转换为详细级别，概要级别，存在级别和删除级别等四个级别，以将其存储在内容存储部分40中。生命单元32监视存储内容的生命周期。内容删除单元33删除删除级别的内容或不太重要的内容，并且使得过滤器单元23登记关于删除的内容的信息。版权所有（C）2006，JPO＆NCIPI

7. 发明专利

JP2004114285A Robotic device and its behavior control method 有权
标题翻译：机器人及其行为控制方法
公开(公告)号：JP2004114285A
公开(公告)日：2004-04-15
申请号：JP2003309262
申请日：2003-09-01
申请人： Sony Corp , ソニー株式会社
发明人： DI PROFIO UGO , FUJITA MASAHIRO , TAKAGI TAKESHI , YOSHIIKE YUKIKO , SHIMOMURA HIDEKI
IPC分类号： A63H11/00 , B25J13/00
摘要： PROBLEM TO BE SOLVED: To realize complicated behaviors by a plurality of behavior modules and to facilitate a recombination of units.
SOLUTION: A situated behavior layer(SBL) 58 is constituted as a tree structure where a plurality of schemas (behavior modules) are hierarchically connected and the independence of each schema is increased to allow each schema to execute independent motions. In this case, a parent schema can define an OR type pattern which gets a child schema to separately execute motions, an AND type pattern which gets a plurality of child schemas to separately execute motions or the connection patterns of child schemas like a SEQUENCE type pattern showing the behavior sequences of a plurality of child schemas and can change behavior patterns which emerge. In addition, a new child schema can be added without rewriting the schema, thereby new motions or functions can be added to the motions of the robotic device.
COPYRIGHT: (C)2004,JPO
摘要翻译：要解决的问题：通过多个行为模块实现复杂的行为并促进单元的重组。解决方案：位置行为层（SBL）58被构造为树结构，其中多个模式（行为模块）被分层连接，并且每个模式的独立性被增加以允许每个模式执行独立的运动。在这种情况下，父模式可以定义一个OR类型模式，该模式获取子模式以单独执行运动，AND类型模式获得多个子模式，以单独执行子模式的运动或连接模式，如SEQUENCE模式显示多个子模式的行为序列，并且可以改变出现的行为模式。此外，可以添加新的子模式而不重写模式，从而可以将新的动作或功能添加到机器人装置的运动。版权所有（C）2004，JPO

8. 发明专利

JP2011018245A Recognition device and method, program, and recording medium 审中-公开
标题翻译：识别装置和方法，程序和记录介质
公开(公告)号：JP2011018245A
公开(公告)日：2011-01-27
申请号：JP2009163192
申请日：2009-07-09
申请人： Sony Corp , ソニー株式会社
发明人： YOSHIIKE YUKIKO , KAWAMOTO KENTA , NODA KUNIAKI , SABE KOTARO
IPC分类号： G06N3/00
摘要： PROBLEM TO BE SOLVED: To properly recognize whether a current node is a learnt internal node or a new internal node to be added, in autonomous learning in changing environment.SOLUTION: A value of a variable N is set to 1. In step S202, time-series information of a length N is acquired. In step S203, a recognizer outputs a node series by use of the Viterbi algorithm based on the time-series information. In step S204, it is decided whether the node series can actually exist or not. When it is decided that the node series cannot actually exist, the node is recognized as an unknown node. When it is decided that the node series can actually exist, entropy is calculated, the value of the variable N is incremented when it is decided that the entropy is a threshold value or above, and the time-series information is extended to a past direction. When it is decided that it is less than the threshold value, it is recognized that the node is a known node.
摘要翻译：要解决的问题：在改变环境中的自主学习中，正确识别当前节点是学习内部节点还是要添加的新内部节点。解决方案：将变量N的值设置为1.在步骤S202中，获取长度为N的时间序列信息。在步骤S203中，识别器基于时间序列信息使用维特比算法输出节点序列。在步骤S204中，确定节点序列是否可以实际存在。当确定节点序列不能实际存在时，节点被识别为未知节点。当确定节点序列可以实际存在时，计算熵，当确定熵是阈值或更高时，变量N的值增加，并且时间序列信息被扩展到过去方向。当判断为小于阈值时，认识到该节点是已知节点。

9. 发明专利

JP2007299366A Learning system and method, recognition device and method, creation device and method, recognition and creation device and method, and program 审中-公开
标题翻译：学习系统和方法，识别装置和方法，创建装置和方法，识别和创建装置及方法及程序
公开(公告)号：JP2007299366A
公开(公告)日：2007-11-15
申请号：JP2006225203
申请日：2006-08-22
申请人： Sony Corp , ソニー株式会社
发明人： KAWAMOTO KENTA , MINAMINO KATSUKI , ITO MASATO , YOSHIIKE YUKIKO , SUZUKI HIROTAKA
IPC分类号： G06N3/08 , G06N3/00
CPC分类号： G06K9/00536 , G06K9/00335 , G06N3/0445 , G06N99/005
摘要： PROBLEM TO BE SOLVED: To continuously recognize and generate data in response to an input from an external environment on the basis of many dynamics by learning the many dynamics. SOLUTION: A learning part 1-4 holds dynamics in one node and performs self-organizational update of a dynamic storage network by a plurality of nodes on the basis of measured time-series data. A recognizing part 1-6 uses the dynamic storage network to recognize the measured time-series data, and a generating part 1-9 uses a node of the dynamic storage network designated by a control signal 1-8 to generate time-series data. The invention is applicable to, for example, a robot. COPYRIGHT: (C)2008,JPO&INPIT
摘要翻译：要解决的问题：通过学习许多动态，通过基于许多动力学来响应来自外部环境的输入来连续地识别和生成数据。解决方案：学习部分1-4在一个节点中保持动态，并且基于测量的时间序列数据，通过多个节点执行动态存储网络的自组织更新。识别部分1-6使用动态存储网络来识别所测量的时间序列数据，并且生成部分1-9使用由控制信号1-8指定的动态存储网络的节点来生成时间序列数据。本发明可应用于例如机器人。版权所有（C）2008，JPO＆INPIT

10. 发明专利

JP2007280054A Learning device, learning method, and program 审中-公开
标题翻译：学习设备，学习方法和程序
公开(公告)号：JP2007280054A
公开(公告)日：2007-10-25
申请号：JP2006105546
申请日：2006-04-06
申请人： Sony Corp , ソニー株式会社
发明人： MINAMINO KATSUKI , ITO MASATO , KAWAMOTO KENTA , YOSHIIKE YUKIKO , SUZUKI HIROTAKA
IPC分类号： G06N3/08 , G06N3/00
CPC分类号： G06N99/005 , G06N3/0445 , G06N3/0454 , G06N3/084
摘要： PROBLEM TO BE SOLVED: To efficiently learn dynamics.
SOLUTION: A winner node determination part 7-2 determines a winner node that is a node corresponding to the dynamics most suitable for an observed time-serial data, out of the plurality of nodes in a network constituted of the plurality of nodes for holding the dynamics. A learning weight determining part 7-3 determines a weight of learning of the dynamics held by the node in every of the nodes, in response to a distance from the winner node. A parameter updating part 7-4 learns self-organizingly the each dynamics of the network by a degree corresponding to the weight of learning, based on the time-serial data. This learning device/method of the present invention is applicable, for example, for a robot.
COPYRIGHT: (C)2008,JPO&INPIT
摘要翻译：要解决的问题：有效地学习动力学。解决方案：胜利者节点确定部分7-2在由多个节点构成的网络中的多个节点中确定作为最适合于观测时间串行数据的动态对应的节点的胜利者节点持有动力。学习权重确定部分7-3响应于与获胜者节点的距离，确定节点在每个节点中所持有的动态学习的权重。参数更新部分7-4基于时间序列数据，以与学习的权重相对应的程度，自组织地学习网络的每个动态。本发明的该学习装置/方法例如适用于机器人。版权所有（C）2008，JPO＆INPIT

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式