Deployment of Dynamics-Based Information Processing Model of Intelligence

基于动力学的情报信息处理模型的部署

• 批准号: 15100002
• 负责人: NAKAMURA Yoshihiko
• 金额: $ 78.12万
• 依托单位: The University of Tokyo
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (S)
• 财政年份: 2003
• 资助国家: 日本
• 起止时间: 2003 至 2007
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-15100002/
• 关键词: Intelligent Robots; Brain-like Information Processing; Somato-Sensory Information; Brain, Neuron; Motion Patterns; Symbolization; Hierarchization; Segmentation; 力学的情報処理; ミメシス; ヒューマノイド; 統計的情報処理; 隠れマルコフモデル; コミュニケーション; 神経筋骨格情報; 知能; 筋骨格モデル; ミ

1. Dynamical Information Processing Theory We have established two methods for designing dynamical information processing mechanisms based on polynomials and physical dynamics systems. We also introduced a parameter representing the plasticity of dynamical systems and developed learning and development model based on the plasticity. A communication model was realized utilizing the attractors of the dynamical system.2. Mathematical Model of Mirror Neurons We established a mathematical model and its computational algorithms for modeling mirror neurons using Hidden Markov Models (HMM). Based on hierarchical HMMs, we realized the abstract ion of behaviors. We also combined the hierarchical mirror neuron model with language analysis system equipped with a common sense database.3. Experimental Study on Perception and Generation of Behaviors through Humanoid Robots We designed and built a humanoid robot that moves in accordance with the gravity. We also developed a method for acquiring motion patterns and controllers by measuring human motions. Finally, we performed several experiments combining the humanoid robot with the dynamical information processing system, language analysis system, and behavior perception and generation system.4. Simulation of Large-Scale Sensor-Motor System through Detailed Human Musculo-Skeletal Models We developed algorithms for estimating the muscle tensions and simulation human motions based on motion capture data. We accelerated the algorithms for detailed human model by employing parallel processing. A large-scale model was developed for simulating the sensor-motor system of the human body and combined with the dynamical information processing system.

1. 动态信息处理理论我们建立了两种基于多项式和物理动力学系统的动态信息处理机制设计方法。我们还引入了表征动力系统可塑性的参数，并建立了基于可塑性的学习和发展模型。利用动力系统的吸引子实现了通信模型。利用隐马尔可夫模型（HMM）建立了镜像神经元建模的数学模型及其计算算法。在分层hmm的基础上，实现了行为的抽象。我们还将分层镜像神经元模型与配备常识数据库的语言分析系统相结合。仿人机器人感知与行为生成实验研究我们设计并制造了一个根据重力运动的仿人机器人。我们还开发了一种通过测量人体运动来获取运动模式和控制器的方法。最后，将仿人机器人与动态信息处理系统、语言分析系统、行为感知与生成系统相结合，进行了实验研究。通过详细的人体肌肉-骨骼模型模拟大规模传感器-运动系统我们开发了基于动作捕捉数据的估计肌肉张力和模拟人体运动的算法。采用并行处理的方法，加快了精细人体模型的求解速度。建立了模拟人体传感-运动系统的大尺度模型，并与动态信息处理系统相结合。

项目成果

力学的情報処理におけるアトラクタ設計法に関する研究-逐次設計による可塑性の導入と階層化設計による大規模化-

动态信息处理中吸引子设计方法研究 - 通过顺序设计引入可塑性，通过分层设计放大 -

• 发表时间: 2005
• 期刊: 日本ロボット学会誌 Vol.23, No.5
• 作者: 岡田 昌史;中村大介;門根秀樹;梶山博史;中村仁彦
• 通讯作者: 中村仁彦

Motion Emergence of Humanoid Robots by an Attractor Design of a Nonlinear Dynamics

非线性动力学吸引子设计的人形机器人运动涌现

• 发表时间: 2005
• 期刊: Proc.of the IEEE International Conference on Robotics and Automation (ICRA'05)
• 作者: Masafumi Okada;Kenta Osato;Yoshihiko Nakamura
• 通讯作者: Yoshihiko Nakamura

Masafumi OKADA, Daisuke NAKAMURA, Yoshihiko NAKAMURA: "On-line and Hierarchical Design Methods of Dynamics Based Information Processing System"Proc. of IEEE/RSJ International Conference on Intelligent Robots and Systems. (CD-ROM). (2003)

Masafumi OKADA、Daisuke NAKAMURA、Yoshihiko NAKAMURA：“基于动力学的信息处理系统的在线和分层设计方法”Proc。

A Fast Online Gait Planning with Boundary Condition Relaxation for Humanoid Robots

• DOI: 10.1109/robot.2005.1570136
• 发表时间: 2005-04
• 期刊: Proceedings of the 2005 IEEE International Conference on Robotics and Automation
• 作者: T. Sugihara;Yoshihiko Nakamura

An omnidirectional vehicle on a basketball

• DOI: 10.1109/icar.2005.1507466
• 发表时间: 2005-07
• 期刊: ICAR '05. Proceedings., 12th International Conference on Advanced Robotics, 2005.
• 作者: T. Endo;Y. Nakamura