基于ARM的六足仿生机器人野外定位系统

为了更好地控制六足仿生机器人适应野外作业环境,针对机器人野外定位问题,提出了一种六足仿生减灾救援机器人无线野外定位系统解决方案,方案以三星S3C2440为硬件平台,以嵌入式linux系统为软件平台,设计了六足仿生机器人野外定位系统。通过GPS全球定位系统进行六足仿生机器人的定位,利用GPRS实现网络通信,并将定位信息传输到终端设备,终端设备通过发送命令的方式控制六足仿生机器人实现相应的动作。实验证明：该系统的稳定性好,可靠性较高,能较好的满足六足仿生减灾救援机器人野外定位的需求。     

Speech feature recognition by profoundly hearing impaired children using a multiple-channel electrotactile speech processor and aided residual hearing

A group of prelinguistically hearing impaired children, between 7 and 11 years of age, were trained in the perception of vowel duration and place, the fricative /s/, and manner of articulation (/m/ vs /b/ and /s/ vs /t/) distinctions, using information provided by a multiple-channel electrotactile aid (Tickle Talker), and through aided hearing. Training was provided in the tactile-plus-aided hearing (TA) and tactile (T) conditions. Speech feature recognition tests were conducted in the TA, T, and aided hearing (A) conditions, during pretraining, training, and post-training phases. Test scores in the TA and T conditions were significantly greater than scores in the A condition for all tests, suggesting that perception of these features was improved when the tactile aid was worn. Test scores in the training and post-training phases were significantly greater than in the pretraining phase, suggesting that the training provided was responsible for the improvement in feature perception. Statistical analyses demonstrated a significant interaction between the main effects of condition and phase, suggesting that training improved perception in the TA and T conditions, but not in the A condition. Post-training and training test scores were similar suggesting that the perceptual skills acquired during training were retained after the removal of training. Recognition of trained features improved for trained, as well as for untrained words.     

Tactile enhancement of auditory and visual speech perception in untrained perceivers

A single pool of untrained subjects was tested for interactions across two bimodal perception conditions: audio-tactile, in which subjects heard and felt speech, and visual-tactile, in which subjects saw and felt speech. Identifications of English obstruent consonants were compared in bimodal and no-tactile baseline conditions. Results indicate that tactile information enhances speech perception by about 10 percent, regardless of which other mode (auditory or visual) is active. However, within-subject analysis indicates that individual subjects who benefit more from tactile information in one cross-modal condition tend to benefit less from tactile information in the other.     

Pitch Shift in Exsomatized Cochlea Observed by Laser Interferometry

Pitch is one of the most important auditory perception characteristics of sound; however, the mechanism underlying the pitch perception of sound is unclear. Although theoretical researches have suggested that perception of virtual pitch is connected with physics in cochlea of inner ear, there is no direct experimental observation of virtual pitch processing in the cochlea. By laser interferometry, we observe shift phenomena of virtual pitch in basilar membrane vibration of exsomatized cochlea, which is consistent with perceptual pitch shift observed in psychoacoustic experiments. This means that the complex mechanical vibration of basilar membrane in cochlea plays an important role in pitch information processing during hearing.     

Perception of acoustic scale and size in musical instrument sounds

There is size information in natural sounds. For example, as humans grow in height, their vocal tracts increase in length, producing a predictable decrease in the formant frequencies of speech sounds. Recent studies have shown that listeners can make fine discriminations about which of two speakers has the longer vocal tract, supporting the view that the auditory system discriminates changes on the acoustic-scale dimension. Listeners can also recognize vowels scaled well beyond the range of vocal tracts normally experienced, indicating that perception is robust to changes in acoustic scale. This paper reports two perceptual experiments designed to extend research on acoustic scale and size perception to the domain of musical sounds: The first study shows that listeners can discriminate the scale of musical instrument sounds reliably, although not quite as well as for voices. The second experiment shows that listeners can recognize the family of an instrument sound which has been modified in pitch and scale beyond the range of normal experience. We conclude that processing of acoustic scale in music perception is very similar to processing of acoustic scale in speech perception.     

Integration of auditory and vibrotactile stimuli: effects of frequency

Perceptual integration of vibrotactile and auditory sinusoidal tone pulses was studied in detection experiments as a function of stimulation frequency. Vibrotactile stimuli were delivered through a single channel vibrator to the left middle fingertip. Auditory stimuli were presented diotically through headphones in a background of 50 dB sound pressure level broadband noise. Detection performance for combined auditory-tactile presentations was measured using stimulus levels that yielded 63% to 77% correct unimodal performance. In Experiment 1, the vibrotactile stimulus was 250 Hz and the auditory stimulus varied between 125 and 2000 Hz. In Experiment 2, the auditory stimulus was 250 Hz and the tactile stimulus varied between 50 and 400 Hz. In Experiment 3, the auditory and tactile stimuli were always equal in frequency and ranged from 50 to 400 Hz. The highest rates of detection for the combined-modality stimulus were obtained when stimulating frequencies in the two modalities were equal or closely spaced (and within the Pacinian range). Combined-modality detection for closely spaced frequencies was generally consistent with an algebraic sum model of perceptual integration; wider-frequency spacings were generally better fit by a Pythagorean sum model. Thus, perceptual integration of auditory and tactile stimuli at near-threshold levels appears to depend both on absolute frequency and relative frequency of stimulation within each modality.     

Recent progress on two-dimensional neuromorphic devices and artificial neural network

Mimicking biological synapses with microelectronic devices is widely considered as the first step in hardware building artificial neuromorphic networks, which is also the basis of brain-inspired neuromorphic computing. Numerous artificial neurons and synapses making up an artificial neuromorphic network have been gained wide attention due to their powerful and efficient data processing capabilities. Recently, artificial synapses, especially memristor-type and transistor-type synapses based on multifarious two-dimensional (2D) materials have been paid much attention. The unique properties of 2D materials make devices perform well in learning ability and power efficiency when mimicking synaptic behaviors, which highlights the feasibility of 2D neuromorphic devices in constructing artificial neuromorphic networks. Herein, the basic structures and principles of biological synapses are introduced, and the definitions of synaptic behaviors in synaptic electronic devices are discussed. Then, the progress of 2D memristor-type and transistor-type neuromorphic devices involving their device architecture, neuromorphic operational mechanism, and promising applications is reviewed. Finally, the future challenges of artificial synaptic devices based on 2D materials are discussed briefly.     

Synthetic vowel studies on cochlear implant patients

Speech perception studies were conducted on three cochlear implant patients to investigate the relative merits of six speech processing schemes for presenting speech information to these patients. Electrical stimuli, described in this article as synthetic vowels, were constructed using tabulated data of formant frequencies of natural vowels. The six schemes differed in the number of formant frequencies encoded on the electrical signal dimension of electrode position, and/or in the range of electrode position used for encoding each formant frequency. Eleven synthetic vowels (i, I, E, ae, a, c, U, u, v, E, D) were used and were presented in a single-interval procedure for absolute identification. Single-formant vowels were used in two of the six schemes, two-formant vowels in three schemes, and three-formant vowels in the remaining scheme. The confusion matrices were subjected to conditional information transmission analysis on the basis of previous psychophysiological findings. Comparisons among the schemes in terms of the analyzed results showed that training, experience, and adaptability to new speech processing schemes were major factors influencing the identification of synthetic vowels. For vowels containing more than one formant, the information about each formant affected the perception of the other formants. In addition, there appeared to be differences between the perceptual processes for vowels containing more than one formant and the processes for single-formant vowels. Taking into consideration the effects of training, experience, and adaptability, the three-formant speech processing scheme appeared, on the basis of perceptual performance comparisons among the six schemes, to be the logical choice for implementation in speech processors for cochlear implant patients.     

N型局部有源忆阻器的神经形态行为

局部有源忆阻器(locally-active memristor,LAM)凭借其高集成度、低功耗和局部有源特性等优点,在神经形态计算领域显示出巨大的潜力.本文提出了一种简单的N型LAM数学模型,通过揭示其非线性动力特性,设计了N型LAM神经元电路.采用Hopf分岔、数值分析等方法定量研究了该电路的动力学行为,成功模拟了多种神经形态行为,包括全或无行为、尖峰、簇发、周期振荡等.并利用该神经元电路结构模拟了生物触觉神经元的频率特性.仿真结果表明:当输入信号幅值低于阈值时,神经元电路输出信号的振荡频率与输入信号强度呈正相关(即兴奋状态),并在阈值处达到最大值.随后,继续增大激励强度,振荡频率则逐渐降低(即保护性抑制状态).最后,设计了N型LAM硬件仿真器,并完成了人工神经元电路的硬件实现,实验结果与仿真结果、理论分析相一致,验证了该N型LAM具备的神经形态行为.     

N型局部有源忆阻器的神经形态行为

局部有源忆阻器(locally-active memristor,LAM)凭借其高集成度、低功耗和局部有源特性等优点,在神经形态计算领域显示出巨大的潜力.本文提出了一种简单的N型LAM数学模型,通过揭示其非线性动力特性,设计了N型LAM神经元电路.采用Hopf分岔、数值分析等方法定量研究了该电路的动力学行为,成功模拟了多种神经形态行为,包括全或无行为、尖峰、簇发、周期振荡等.并利用该神经元电路结构模拟了生物触觉神经元的频率特性.仿真结果表明:当输入信号幅值低于阈值时,神经元电路输出信号的振荡频率与输入信号强度呈正相关(即兴奋状态),并在阈值处达到最大值.随后,继续增大激励强度,振荡频率则逐渐降低(即保护性抑制状态).最后,设计了N型LAM硬件仿真器,并完成了人工神经元电路的硬件实现,实验结果与仿真结果、理论分析相一致,验证了该N型LAM具备的神经形态行为.     

Target Classification Method of Tactile Perception Data with Deep Learning

In order to improve the accuracy of manipulator operation, it is necessary to install a tactile sensor on the manipulator to obtain tactile information and accurately classify a target. However, with the increase in the uncertainty and complexity of tactile sensing data characteristics, and the continuous development of tactile sensors, typical machine-learning algorithms often cannot solve the problem of target classification of pure tactile data. Here, we propose a new model by combining a convolutional neural network and a residual network, named ResNet10-v1. We optimized the convolutional kernel, hyperparameters, and loss function of the model, and further improved the accuracy of target classification through the K-means clustering method. We verified the feasibility and effectiveness of the proposed method through a large number of experiments. We expect to further improve the generalization ability of this method and provide an important reference for the research in the field of tactile perception classification.     

基于触滑觉感知的智能假肢抓握控制方法研究

【】触觉和滑觉感知功能是仿生假手不可或缺的感知功能。本文在原有智能假肢的基础上,通过对触滑觉感知方法的研究,包括触滑觉传感器的选型、后续测量电路的设计和测点布局分析等,开发了一种可以同时实现触觉和滑觉感知的智能假肢。进而,通过模糊逻辑控制方法的引入,实现了假肢的可靠抓握功能及抓握保持过程中的自适应响应控制。实验结果表明,该智能假肢可以实现假肢抓握的稳定控制,并在被抓握物体产生滑动时进行精准快速的自适应响应控制。     

Detectability of words and nonwords in two kinds of noise

Recent models of speech perception emphasize the possibility of interactions among different processing levels. There is evidence that the lexical status of an utterance (i.e., whether it is a meaningful word or not) may influence earlier stages of perceptual analysis. To test how far down such "top-down" influences might penetrate, an investigation was conducted to determine whether there is a difference in detectability of words and nonwords masked by amplitude-modulated or unmodulated broadband noise. The results were negative, suggesting either that the stages of perceptual analysis engaged in the detection task are impermeable to lexical top-down effects, or that the lexical level was not sufficiently activated to have any facilitative effect on perception.     

Signal feature recognition based on lightwave neuromorphic signal processing

We developed a hybrid analog/digital lightwave neuromorphic processing device that effectively performs signal feature recognition. The approach, which mimics the neurons in a crayfish responsible for the escape response mechanism, provides a fast and accurate reaction to its inputs. The analog processing portion of the device uses the integration characteristic of an electro-absorption modulator, while the digital processing portion employ optical thresholding in a highly Ge-doped nonlinear loop mirror. The device can be configured to respond to different sets of input patterns by simply varying the weights and delays of the inputs. We experimentally demonstrated the use of the proposed lightwave neuromorphic signal processing device for recognizing specific input patterns.     

基于忆容器件的神经形态计算研究进展

人工智能的快速发展需要人工智能专用硬件的快速发展,受人脑存算一体、并行处理启发而构建的包含突触与神经元的神经形态计算架构,可以有效地降低人工智能中计算工作的能耗.记忆元件在神经形态计算的硬件实现中展现出巨大的应用价值;相比传统器件,用忆阻器构建突触、神经元能极大地降低计算能耗,然而在基于忆阻器构建的神经网络中,更新、读取等操作存在由忆阻电压电流造成的系统性能量损失.忆容器作为忆阻器衍生器件,被认为是实现低耗能神经网络的潜在器件,引起国内外研究者关注.本文综述了实物/仿真忆容器件及其在神经形态计算中的最新进展,主要包括目:前实物/仿真忆容器原理与特性,代表性的忆容突触、神经元及神经形态计算架构,并通过总结近年来忆容器研究所取得的成果,对当前该领域面临的挑战及未来忆容神经网络发展的重点进行总结与展望.     

基于PSO算法的RoboCup2D机器人研究

对机器人体系结构、动作学习及行为的组织方式进行了研究,以演化计算为基本方法,以RoboCup2D为平台,设计了基于PSO算法的足球机器人的体系结构,解决感知、动作、和规划问题;在训练环境下,形成感知规则,优化感知相关参数,得到对信息高效快速的感知方法,并根据指定的粒度、功能、参数,对RoboCup2D机器人的原子动作进行了组合优化,得到一组带参数和执行效果描述的粒子动作;最后在赛场环境和任务驱动下,搜索粒子动作并进行组织规划,得到完成特定任务的机器人行为;RoboCup2D仿真实验表明,演化计算方法不仅能利用原子动作进行组合优化,得到适应于不同条件的粒子动作,而且能通过其在线搜索粒子动作,动态组成机器人行为;基于演化计算的足球机器人能更好地完成跑位、截球、带球、传球等任务,具有更强的适应性。     

Non-linearity of the interactions between bottom-up and top-down signals in multistable visual perception

Summary Both bottom-up sensory information and top-down influences contribute to the perception processes. We studied the perceptual alternations of a multistable ambiguous pattern. We observed that it is possible to interfere on the process of the perception alternance by means of subliminal visual stimuli, which either contrast or second the previous perception. We investigated also the effect of the top-down volitional factor on the perceptual alternation. By using a combination of such top-down factor and bottom-up stimulation, we ascertained that a non-linear type of interaction occurs between the two above factors. Paper presented at the International Workshop ?Fluctuations in Physics and Biology: Stochastic Resonance, Signal Processing and Related Phenomena?, Elba, 5–10 June 1994.     

Speech perception studies using a multichannel electrotactile speech processor, residual hearing, and lipreading

Three studies are reported on the speech perception of normally hearing and hearing-impaired adults using combinations of visual, auditory, and tactile input. In study 1, mean scores for four normally hearing subjects showed that addition of tactile information, provided through the multichannel electrotactile speech processor, to either audition alone (300-Hz low-pass-filtered speech) or lipreading plus audition resulted in significant improvements in phoneme and word discrimination scores. Information transmission analyses demonstrated the effectiveness of the tactile aid in providing cues to duration, F1 and F2 features for vowels, and manner of articulation features for consonants, especially features requiring detection and discrimination of high-frequency information. In study 2, six different cutoff frequencies were used for a low-pass-filtered auditory signal. Mean scores for vowel and consonant identification were significantly higher with the addition of tactile input to audition alone at each cutoff frequency up to 1500 Hz. The mean speechtracking rate was also significantly increased by the additional tactile input up to 1500 Hz. Study 3 examined speech discrimination of three hearing-impaired adults. Additional information available through the tactile aid was shown to improve speech discrimination scores; however, the degree of increase was inversely related to the level of residual hearing. Results indicate that the electrotactile aid may be useful for patients with little residual hearing and for the severely to profoundly hearing impaired, who could benefit from the high-frequency information presented through the tactile modality, but unavailable through hearing aids.     

Four channels mediate the mechanical aspects of touch

Although previous physiological and anatomical experiments have identified four afferent fiber types (PC, RA, SA II, and SA I) in glabrous (nonhairy) skin of the human somatosensory periphery, only three have been shown to mediate tactile (mechanoreceptive) sensation. Psychophysical evidence that four channels (P, NP I, NP II, and NP III) do, indeed, participate in the perceptual process is presented. In a series of experiments involving selective masking of the various channels, modification of the skin-surface temperature, and testing cutaneous sensitivity down to very low-vibratory frequencies, the fourth psychophysical channel (NP III) is defined. Based on these experiments and previous work from our laboratory, it is concluded that the four channels work in conjunction at threshold to create an operating range for the perception of vibration that extends from at least 0.4 to greater than 500 Hz. Each of the four channels appears to mediate specific portions of the overall threshold-frequency characteristic. Selection of appropriate neural-response criteria from previously published physiological data and correlation of their derived frequency characteristics with the four psychophysical channels indicates that each channel has its own physiological substrate: P channel and PC fibers, NP I channel and RA fibers, NP II channel and SA II fibers, and NP III channel and SA I fibers. These channels partially overlap in their absolute sensitivities, making it likely that suprathreshold stimuli may activate two or more of the channels at the same time. Thus the perceptual qualities of touch may be determined by the combined inputs from four channels.     

Quantification of Gloss Perception as a Function of Stimulus Duration

The mechanism and temporal characteristics of gloss perception are not entirely clear. In addition, the formulation for predicting gloss perception from photometric values has not been established. In the present study, we conducted an experiment to measure several temporal characteristics of gloss perception in order to clarify the mechanism. All stimuli were rendered as computer graphics with Phong and Lambert models to provide gloss perception to human observers. We measured perceptual glossiness with a magnitude estimation method and perceptual diffuse/specular reflectance of test stimuli with a matching method under several stimulus conditions, such as reflectance coefficients and stimulus duration. The results showed that the perceptual specular component and perceptual glossiness increase with decreasing stimulus duration. Finally, we proposed a formulation to predict perceptual glossiness as a function of stimulus duration.