Initial conditions, or initial resetting play a fundamental role in all kinds of iterative learning control methods. In this note, we study five different disclose the inherent relationship between each condition and corresponding convergence (or boundedness) property. The method under consideration is based on Lyapunov theory, which suitable for plants with time-varying parametric uncertainties local Lipschitz nonlinearities.

In this paper, we propose a role adaptation method for human-robot shared control. Game theory is employed fundamental analysis of two-agent system. An law developed such that the robot able to adjust its own according human's intention lead or follow, which inferred through measured interaction force. absence human forces, adaptive scheme allows take and complete task by itself. On other hand, when persistently exerts strong forces signal an unambiguous intent lead, yields becomes follower....

In this paper, we propose a framework to analyze the interactive behaviors of human and robot in physical interactions.Game theory is employed describe system under study, policy iteration adopted provide solution Nash equilibrium.The human's control objective estimated based on measured interaction force, it used adapt robot's such that human-robot coordination can be achieved.The validity proposed method verified through rigorous proof experimental studies.

In this paper, we address two challenging problems in unsupervised subspace learning: 1) how to automatically identify the feature dimension of learned (i.e., automatic learning) and 2) learn underlying presence Gaussian noise robust learning). We show that these can be simultaneously solved by proposing a new method [(called principal coefficients embedding (PCE)]. For given data set , PCE recovers clean from learns global reconstruction relation . By preserving into an -dimensional space,...

A problem that hinders good performance of general gait recognition systems is the appearance features gaits are more affected-prone by views than identities, especially when walking direction probe different from register gait. This cannot be solved traditional projection learning methods because these can learn only one matrix, and thus for same subject, it transfer cross-view into similar ones. paper presents an innovative method to overcome this aligning energy images (GEIs) across with...

Video Temporal Grounding (VTG), which aims to ground target clips from videos (such as consecutive intervals or disjoint shots) according custom language queries (e.g., sentences words), is key for video browsing on social media. Most methods in this direction develop task-specific models that are trained with type-specific labels, such moment retrieval (time interval) and highlight detection (worthiness curve), limits their abilities generalize various VTG tasks labels. In paper, we propose...

This paper proposes a biologically plausible network architecture with spiking neurons for sequence recognition. is unified and consistent system functional parts of sensory encoding, learning, decoding. the first systematic model attempting to reveal neural mechanisms considering both upstream downstream together. The whole temporal framework, where precise timing spikes employed information processing cognitive computing. Experimental results show that competent perform recognition, being...

In this note, a repetitive learning control (RLC) approach is proposed to deal with periodic tracking tasks for nonlinear dynamical systems nonparametric uncertainties. We address two fundamental issues associated the methodology: The existence of solution, and convergence property. Applying theorem neutral differential difference equation, using Lyapunov-Krasovskii functional, solution can be proven rigorously. A further extension RLC cascade also explored

We present adaptive admittance control of a robotic manipulator, with uncertain dynamic parameters, operating in constrained task space. To provide compliance to external forces, we generate differentiable reference trajectory that remains the Then, backstepping control, based on time-varying asymmetric Barrier Lyapunov Function (BLF), is designed achieve tracking while guaranteeing constraint satisfaction. The improved BLF-based renders entire space positively invariant. Despite transient...

One of the important topics in study robotic cognition is to enable robot perceive, plan, and react situations a real-world environment. We present novel angle on this subject, by integrating active navigation with sequence learning. propose neuro-inspired cognitive model which integrates mapping ability entorhinal cortex (EC) episodic memory hippocampus perform more versatile tasks. The EC layer modeled 3-D continuous attractor network structure build map recurrent spiking neural store...

Address event representation (AER) image sensors represent the visual information as a sequence of events that denotes luminance changes scene. In this paper, we introduce feature extraction method for AER based on probability theory, namely, bag (BOE). The proposed approach represents each object joint distribution concurrent events, and corresponds to unique activated pixel sensor. advantages BOE include: 1) it is statistical learning has good interpretability in mathematics; 2) can...

Spiking neural networks (SNNs) have shown great potential as a solution for realizing ultralow-power consumption on neuromorphic hardware, but obtaining deep SNNs is still challenging problem. Existing network conversion methods can effectively obtain from the trained convolutional (CNNs) with little performance loss, however, high-precision weights in converted would take up high-storage space nonamicable to limited memory resources. To tackle this problem, we analyze relationship between...

Gait recognition can be used in person identification and re-identification by itself or conjunction with other biometrics. Although gait has both spatial temporal attributes, it been observed that decoupling feature better exploit the on fine-grained level. However, spatial-temporal correlations of video signals are also lost process. Direct 3D convolution approaches retain such correlations, but they introduce unnecessary interferences. Instead common solutions, this paper proposes an...

Event cameras are asynchronous and neuromorphically inspired visual sensors, which have shown great potential in object tracking because they can easily detect moving objects. Since event output discrete events, inherently suitable to coordinate with Spiking Neural Network (SNN), has a unique event-driven computation characteristic energy-efficient computing. In this paper, we tackle the problem of event-based by novel architecture discriminatively trained SNN, called Convolutional Tracking...

Spiking neural networks (SNNs) operating with asynchronous discrete events show higher energy efficiency sparse computation. A popular approach for implementing deep SNNs is artificial network (ANN)-SNN conversion combining both efficient training of ANNs and inference SNNs. However, the accuracy loss usually nonnegligible, especially under few time steps, which restricts applications SNN on latency-sensitive edge devices greatly. In this article, we first identify that such performance...

Polymer nanocomposites consist of a polymer matrix and fillers with at least one dimension below 100 nanometers (nm) [L. Schadler et al., Jom 59(3), 53–60 (2007)]. A key challenge in constructing an effective data resource for is building consistent, coherent, clear representation all relevant parameters their interrelationships. The must address (1) representing, saving, accessing the (e.g., schema used such as database management system), (2) contribution uploading MS Excel template file...

In this article, we present a systematic computational model to explore brain-based computation for object recognition. The extracts temporal features embedded in address-event representation (AER) data and discriminates different objects by using spiking neural networks (SNNs). We use multispike encoding extract contained the AER data. These patterns are then learned through tempotron learning rule. presented is consistently implemented framework, where precise timing of spikes considered...

Spiking neural networks (SNNs) have increasingly drawn massive research attention due to biological interpretability and efficient computation. Recent achievements are devoted utilizing the surrogate gradient (SG) method avoid dilemma of non-differentiability spiking activity directly train SNNs by backpropagation. However, fixed width SG leads vanishing mismatch problems, thus limiting performance trained SNNs. In this work, we propose a novel perspective unlock limitation SG, called...