Developmental word grounding through a growing neural network with a humanoid robot.

This paper presents an unsupervised approach of integrating speech and visual information without using any prepared data. The approach enables a humanoid robot, Incremental Knowledge Robot 1 (IKR1), to learn word meanings. The approach is different from most existing approaches in that the robot learns online from audio-visual input, rather than from stationary data provided in advance. In addition, the robot is capable of learning incrementally, which is considered to be indispensable to lifelong learning. A noise-robust self-organized growing neural network is developed to represent the topological structure of unsupervised online data. We are also developing an active-learning mechanism, called "desire for knowledge," to let the robot select the object for which it possesses the least information for subsequent learning. Experimental results show that the approach raises the efficiency of the learning process. Based on audio and visual data, they construct a mental model for the robot, which forms a basis for constructing IKRI's inner world and builds a bridge connecting the learned concepts with current and past scenes.     

Model-based learning for mobile robot navigation from the dynamicalsystems perspective

This paper discusses how a behavior-based robot can construct a "symbolic process" that accounts for its deliberative thinking processes using models of the environment. The paper focuses on two essential problems; one is the symbol grounding problem and the other is how the internal symbolic processes can be situated with respect to the behavioral contexts. We investigate these problems by applying a dynamical system's approach to the robot navigation learning problem. Our formulation, based on a forward modeling scheme using recurrent neural learning, shows that the robot is capable of learning grammatical structure hidden in the geometry of the workspace from the local sensory inputs through its navigational experiences. Furthermore, the robot is capable of generating diverse action plans to reach an arbitrary goal using the acquired forward model which incorporates chaotic dynamics. The essential claim is that the internal symbolic process, being embedded in the attractor, is grounded since it is self-organized solely through interaction with the physical world. It is also shown that structural stability arises in the interaction between the neural dynamics and the environmental dynamics, which accounts for the situatedness of the internal symbolic process, The experimental results using a mobile robot, equipped with a local sensor consisting of a laser range finder, verify our claims.     

Controlling a robot manipulator with fuzzy voice commands using a probabilistic neural network

Natural language commands are generated by intelligent human beings. As a result, they contain a lot of information. Therefore, if it is possible to learn from such commands and reuse that knowledge, it will be a very efficient process. In this paper, learning from such information rich voice commands for controlling a robot is studied. First, new concepts of fuzzy coach-player system and sub-coach are proposed for controlling robots with natural language commands. Then, the characteristics of the subjective human decision making process are discussed and a Probabilistic Neural Network (PNN) based learning method is proposed to learn from such commands and to reuse the acquired knowledge. Finally, the proposed concept is demonstrated and confirmed with experiments conducted using a PA-10 redundant manipulator.     

Q-学习及其在智能机器人局部路径规划中的应用研究

强化学习一词来自于行为心理学,这门学科把行为学习看成反复试验的过程,从而把环境状态映射成相应的动作．在设计智能机器人过程中,如何来实现行为主义的思想、在与环境的交互中学习行为动作？ 文中把机器人在未知环境中为躲避障碍所采取的动作看作一种行为,采用强化学习方法来实现智能机器人避碰行为学习．Ｑ－学习算法是类似于动态规划的一种强化学习方法,文中在介绍了Ｑ－学习的基本算法之后,提出了具有竞争思想和自组织机制的Ｑ－学习神经网络学习算法;然后研究了该算法在智能机器人局部路径规划中的应用,在文中的最后给出了详细的仿真结果     

Development of an Intelligent Wheelchair Using Computer Graphics Animation and Simulation

A new robot simulator JC-1 is used as a control software development tool in a project in progress where an intelligent wheelchair for a blind user is being developed. The intelligent wheelchair is planned to be able to fulfill simple symbolic commands like "follow wall" or "follow object" and using the JC-1 simulator an evaluation team which includes e.g. the user, a rehabilitation engineer and a software engineer, can check control algorithms and user interface routines before constructing a real wheelchair prototype. The JC-1 simulator models the environment using simplified boundary- representation where objects, robot sensors and actuators are presented as symbolic objects in the graphics data-base of the simulator. In the JC-1 simulator a robot controller under development controls the motion of the graphical model of the robot while simulator commands or other robot controllers can be used to control the movement of disturbing obstacles. Computer graphics animation and simulation help to find fundamental design errors at an early design stage and as this paper suggests, enable the user of the final product to take part in to the designing process of the robot controller. Benefits and difficulties of using computer graphics simulation in the wheelchair development process are discussed.     

基于强化学习的壮语词性标注

目前壮语智能信息处理研究处于起步阶段,缺乏自动词性标注方法.针对壮语标注语料匮乏、人工标注费时费力而机器标注性能较差的现状,提出一种基于强化学习的壮语词性标注方法.依据壮语的文法特点和中文宾州树库符号构建标注词典,通过依存句法分析融合语义特征,并以长短期记忆网络为策略网络,利用循环记忆完善部分观测信息.在此基础上,引入强化学习框架,将目标词性作为环境反馈,通过特征学习不断逼近目标真实值.实验结果表明,该方法可缓解词性标注模型对训练语料库的依赖,能够快速扩大壮语标注词典的规模,实现壮语词性的自动标注.     

Real-time Unsupervised Segmentation of human whole-body motion and its application to humanoid robot acquisition of motion symbols

An interactive loop between motion recognition and motion generation is a fundamental mechanism for humans and humanoid robots. We have been developing an intelligent framework for motion recognition and generation based on symbolizing motion primitives. The motion primitives are encoded into Hidden Markov Models (HMMs), which we call “motion symbols”. However, to determine the motion primitives to use as training data for the HMMs, this framework requires a manual segmentation of human motions. Essentially, a humanoid robot is expected to participate in daily life and must learn many motion symbols to adapt to various situations. For this use, manual segmentation is cumbersome and impractical for humanoid robots. In this study, we propose a novel approach to segmentation, the Real-time Unsupervised Segmentation (RUS) method, which comprises three phases. In the first phase, short human movements are encoded into feature HMMs. Seamless human motion can be converted to a sequence of these feature HMMs. In the second phase, the causality between the feature HMMs is extracted. The causality data make it possible to predict movement from observation. In the third phase, movements having a large prediction uncertainty are designated as the boundaries of motion primitives. In this way, human whole-body motion can be segmented into a sequence of motion primitives. This paper also describes an application of RUS to AUtonomous Symbolization of motion primitives (AUS). Each derived motion primitive is classified into an HMM for a motion symbol, and parameters of the HMMs are optimized by using the motion primitives as training data in competitive learning. The HMMs are gradually optimized in such a way that the HMMs can abstract similar motion primitives. We tested the RUS and AUS frameworks on captured human whole-body motions and demonstrated the validity of the proposed framework.     

自组织映射神经网络量化机器人强化学习方法研究

强化学习一词来自于行为心理学，这门学科把行为学习看成反复试验的过程，从而把环境状态映射成相应的动作。在设计智能机器人过程中，如何来实现行为主义的思想，在与环境的交互中学习行为动作？文中把机器人在未知环境中为躲避障碍所采取的动作看作一种行为，采用强化学习方法来实现智能机器人避碰行为学习。为了提高机器人学习速度，在机器人局部路径规划中的状态空量化就显得十分重要。本文采用自组织映射网络的方法来进行空间的量化。由于自组织映射网络本身所具有的自组织特性，使得它在进行空间量化时就能够较好地解决适应性灵活性问题，本文在对状态空间进行自组织量化的基础方法上，采用强化学习。解决了机器人避碰行为的学习问题，取得了满意的学习结果。     

Goal emulation and planning in perceptual space using learned affordances

In this paper, we show that through self-interaction and self-observation, an anthropomorphic robot equipped with a range camera can learn object affordances and use this knowledge for planning. In the first step of learning, the robot discovers commonalities in its action-effect experiences by discovering effect categories. Once the effect categories are discovered, in the second step, affordance predictors for each behavior are obtained by learning the mapping from the object features to the effect categories. After learning, the robot can make plans to achieve desired goals, emulate end states of demonstrated actions, monitor the plan execution and take corrective actions using the perceptual structures employed or discovered during learning. We argue that the learning system proposed shares crucial elements with the development of infants of 7–10 months age, who explore the environment and learn the dynamics of the objects through goal-free exploration. In addition, we discuss goal emulation and planning in relation to older infants with no symbolic inference capability and non-linguistic animals which utilize object affordances to make action plans.     

Manipulation planning using learned symbolic state abstractions

We present an approach for planning robotic manipulation tasks that uses a learned mapping between geometric states and logical predicates. Manipulation planning, because it requires task-level and geometric reasoning, requires such a mapping to convert between the two. Consider a robot tasked with putting several cups on a tray. The robot needs to find positions for all the objects, and may need to nest one cup inside another to get them all on the tray. This requires translating back and forth between symbolic states that the planner uses, such as stacked (cup1,cup2), and geometric states representing the positions and poses of the objects. We learn the mapping from labelled examples, and importantly learn a representation that can be used in both the forward (from geometric to symbolic) and reverse directions. This enables us to build symbolic representations of scenes the robot observes, but also to translate a desired symbolic state from a plan into a geometric state that the robot can achieve through manipulation. We also show how such a mapping can be used for efficient manipulation planning: the planner first plans symbolically, then applies the mapping to generate geometric positions that are then sent to a path planner.     

INTELLIGENT SENSOR FUSION AND LEARNING FOR AUTONOMOUS ROBOT NAVIGATION

ABSTRACT

This paper presents the design and implementation of an autonomous robot navigation system for intelligent target collection in dynamic environments. A feature-based multi-stage fuzzy logic (MSFL) sensor fusion system is developed for target recognition, which is capable of mapping noisy sensor inputs into reliable decisions. The robot exploration and path planning are based on a grid map oriented reinforcement path learning system (GMRPL), which allows for long-term predictions and path adaptation via dynamic interactions with physical environments. In our implementation, the MSFL and GMRPL are integrated into subsumption architecture for intelligent target-collecting applications. The subsumption architecture is a layered reactive agent structure that enables the robot to implement higher-layer functions including path learning and target recognition regardless of lower-layer functions such as obstacle detection and avoidance. The real-world application using a Khepera robot shows the robustness and flexibility of the developed system in dealing with robotic behaviors such as target collecting in the ever-changing physical environment.     

基于云平台的智能语音交互机器人设计

现有的语音交互机器人多采用用户提问、机器人回答的单向交流方式,人机交互的智能性和灵活性较差。本文研究运用树莓派(Raspberry Pi)计算机和配套的语音板作为硬件载体,融合语音唤醒、语音识别、语音合成、自然语言处理等人工智能技术,调用科大讯飞开放云平台、在线图灵机器人,搭建一种基于云平台的智能语音交互机器人系统,并结合自主开发的本地知识库和问题库,使智能语音交互机器人能够根据不同环境与任务需求实现双向互动交流,实现由机器人采集信息和交流反馈,以提供高适应性的无接触人机语音交互服务。     

Learning proactive behavior for interactive social robots

Learning human–robot interaction logic from example interaction data has the potential to leverage “big data” to reduce the effort and time spent on designing interaction logic or crafting interaction content. Previous work has demonstrated techniques by which a robot can learn motion and speech behaviors from non-annotated human–human interaction data, but these techniques only enable a robot to respond to human-initiated inputs, and do not enable the robot to proactively initiate interaction. In this work, we propose a method for learning both human-initiated and robot-initiated behavior for a social robot from human–human example interactions, which we demonstrate for a shopkeeper interacting with a customer in a camera shop scenario. This was achieved by extending an existing technique by (1) introducing a concept of a customer yield action, (2) incorporating interaction history, represented by sequences of discretized actions, as inputs for training and generating robot behavior, and (3) using an “attention mechanism” in our learning system for training robot behaviors, that learns which parts of the interaction history are more important for generating robot behaviors. The proposed method trains a robot to generate multimodal actions, consisting of speech and locomotion behaviors. We compared this study with the previous technique in two ways. Cross-validation on the training data showed higher social appropriateness of predicted behaviors using the proposed technique, and a user study of live interaction with a robot showed that participants perceived the proposed technique to produce behaviors that were more proactive, socially-appropriate, and better in overall quality.     

短视频不良内容智能检测平台研究与设计

研究并设计了一个面向短视频不良内容的实时检测平台。该平台研究的核心在于分层筛选系统,通过基于短视频外围核心参数构建的深度学习筛选模型完成海量筛选,再将检测出的可疑不良短视频传递给基于深度学习的短视频内容识别引擎进行重点甄别,提出融合自然语言处理、计算机视觉、语音识别、机器学习等的短视频智能实时检测平台框架。     

基于智能语音的翻译机器人自动化控制系统设计

为提升自动控制效果,加快翻译速率,设计基于智能语音的翻译机器人自动化控制系统。采集外界智能语音信号,利用A/D转换器得到数字信号,启动语音唤醒模块激活翻译机器人,听写模式识别复杂语音信号,命令模式识别简单语音信号,得到语言文本识别结果,通过深度学习关键词检测方法提取关键词作为翻译机器人的自动化控制指令,通过单片机识别自动化控制指令。实验结果表明,该系统可有效采集外界智能语音信号,提取智能语音信号的关键词,完成翻译机器人自动化控制。     

Maximum entropy direct models for speech recognition

Traditional statistical models for speech recognition have mostly been based on a Bayesian framework using generative models such as hidden Markov models (HMMs). This paper focuses on a new framework for speech recognition using maximum entropy direct modeling, where the probability of a state or word sequence given an observation sequence is computed directly from the model. In contrast to HMMs, features can be asynchronous and overlapping. This model therefore allows for the potential combination of many different types of features, which need not be statistically independent of each other. In this paper, a specific kind of direct model, the maximum entropy Markov model (MEMM), is studied. Even with conventional acoustic features, the approach already shows promising results for phone level decoding. The MEMM significantly outperforms traditional HMMs in word error rate when used as stand-alone acoustic models. Preliminary results combining the MEMM scores with HMM and language model scores show modest improvements over the best HMM speech recognizer.     

Mutual learning of multi-consciousness agents,including an ego for an autonomous vehicle

In this paper, we propose a multi-agent learning system for the control of an intelligent robot, based on a model of the human consciousnesses, including the ego. We pay attention to the intelligent learning processes of human beings. We try to give a robot a high learning ability by modeling the roles of the human consciousnesses, including the ego. In most ordinary methods, the instructions for learning are given from outside the system only. In the proposed method, the instructions are given not only from outside, but also from inside (from other agents in the system). Therefore, the robot can learn efficiently because it has more instructions than usual. The learning is also more flexible, since an agent learns by instructions from other agents while the learning agent and one of the instructing agents exchange roles according to changes in the environment. We experimentally verified that the proposed method is efficient by using an actual robot.