Constructing topic hierarchies from the data automatically can help us better understand the contents and structure of information and benefit many applications in security informatics. The existing topic hierarchy co...
详细信息
ISBN:
(纸本)9781467384940
Constructing topic hierarchies from the data automatically can help us better understand the contents and structure of information and benefit many applications in security informatics. The existing topic hierarchy construction methods either need to specify the structure manually, or are not robust enough for sparse and noisy social media data such as microblog. In this paper, we propose an approach to automatically construct topic hierarchies from microblog data in a bottom up manner. We detect topics first and then build the topic structure based on a tree combination method. We conduct a preliminary empirical study based on the Weibo data. The experimental results show that the topic hierarchies generated by our method provide meaningful results.
This paper is concerned with data-driven methods for virtual reference controller design of high-order nonlinear systems via neural network. Virtual reference feedback tuning (VRFT) is a one-shot direct data-based met...
详细信息
ISBN:
(纸本)9781479919611
This paper is concerned with data-driven methods for virtual reference controller design of high-order nonlinear systems via neural network. Virtual reference feedback tuning (VRFT) is a one-shot direct data-based method to design controller of linear or nonlinear systems. In this paper, we recall the model reference control problem of high-order nonlinear systems and design a new objective function of VRFT. In ideal conditions, the two problems are demonstrated to have the same solution. For the first time, we prove that the value of the optimization problem for model reference control is bounded by that of the objective function of VRFT. A three-layer neural network is employed as a general approximator of the designed controller and two simulations are given to verify the validity of our method.
Jellyfish-inspired jet propulsion and robotic implementation have drawn much attention among the scientific community. In this paper, we report a novel robotic jellyfish synthesizing mechanical structure drive and bar...
详细信息
ISBN:
(纸本)9781467396769
Jellyfish-inspired jet propulsion and robotic implementation have drawn much attention among the scientific community. In this paper, we report a novel robotic jellyfish synthesizing mechanical structure drive and barycenter adjustment mechanism. The robotic jellyfish relies on six-bar linkage mechanisms as actuators to realize the contraction-relaxation jet propulsion motion. To imitate the real jellyfish and to reduce drag, the robotic jellyfish is comprised of a streamlined head, a cavity shell, and a latex skin is enwraped around the drive units. To achieve free switch between different motion modes and transform attitude arbitrarily, a barycenter adjustment mechanism is embedded in the robot body. Through the upward, downward, balancing, as well as leaning-down motion of clump weights, the center of gravity of the robot can enwrap its center of buoyancy. Experimental tests on the actual robotic jellyfish verify the great 3D swimming ability.
In this paper, a novel robust adaptive control scheme is proposed for a lower limb rehabilitation robot designed by our laboratory. The proposed control strategy is based on the radial basis function (RBF) neural netw...
详细信息
ISBN:
(纸本)9781467396769
In this paper, a novel robust adaptive control scheme is proposed for a lower limb rehabilitation robot designed by our laboratory. The proposed control strategy is based on the radial basis function (RBF) neural networks and the parameters of the system dynamics are unknown. The weights of the RBF neural networks are updated by an adaptive law according to the Lyapunov stability analysis. The robustness against possible variations of the system dynamics and the external disturbance are considered in the control design. The proposed control strategy can not only avoid the complex procedure of system parameters identification, but also guarantee high robustness, small trajectory tracking errors and the assistance of the patient's voluntary participation. Using this control algorithm, the robot can regulate its exerted torque to adapt to the patient's active torque in real time during rehabilitation. The effectiveness of our control method is demonstrated by a simulation.
Reinforcement learning has provided an efficient approach to solve the optimal control of some complicated systems. In this paper we resort to the idea of off-policy scheme and propose a complete-model-free algorithm ...
详细信息
ISBN:
(纸本)9781467371902
Reinforcement learning has provided an efficient approach to solve the optimal control of some complicated systems. In this paper we resort to the idea of off-policy scheme and propose a complete-model-free algorithm to the continuous-time nonlinear optimal control problems. The novel algorithm consists of two neural networks to approximate the value and policy and adapts them with continuous tuning laws. In addition, experience replay technique is employed so that both the instantaneous observations and the past data are utilized. The convergence to the optimal solutions are guaranteed under the Lyapunov analysis. A nonlinear system is simulated to test the learning performance.
This paper aims at dynamic modeling and control of a new upper-limb rehabilitation robot which has a parallel structure. Dynamic modeling of parallel robot is a complicated problem, and the dynamics and voluntary forc...
详细信息
ISBN:
(纸本)9781479918096
This paper aims at dynamic modeling and control of a new upper-limb rehabilitation robot which has a parallel structure. Dynamic modeling of parallel robot is a complicated problem, and the dynamics and voluntary force of the patient arm increase the difficulty of dynamic analysis and control in rehabilitation training. The novelties of this study are: (1) dynamics of the robot and the patient are considered together, and this human-robot interaction system is modeled as a redundantly actuated closed-chain system (2 DOFs, 4 active joints); (2) the system dynamics are derived in workspace using a new method based on the dynamics of its three serial open-chain branches, and both kinematic constrains and interaction forces are considered during the derivation. Compared with the other two previous methods reviewed in this paper, the proposed method is easier to derive, more computationally efficient, and it can be used in both redundant and non-redundant cases. Besides, a model based PD-computed torque controller is designed and the simulation of passive training task along a circular path is presented to prove the effectiveness of this method.
Group consensus has both positive and negative communication weights, which is an extension to traditional consensus problem. Additionally, distributed event-triggered control has advantages over periodic control cons...
详细信息
Group consensus has both positive and negative communication weights, which is an extension to traditional consensus problem. Additionally, distributed event-triggered control has advantages over periodic control considering energy consumption and communication constraints. Thus, it is important to study group consensus using event-triggered control. Moreover, by calculating the maximum and minimum of the corresponding parameters, we can simplify the event-triggered function. The implementation will validate the effectiveness of our distributed control protocol.
This paper presents a novel jelly sh-inspired swimming robot whose mechanical design, control algorithm, and motion analysis are discussed and manifested in detail. In consideration of few research on robotic jelly sh...
详细信息
This paper presents a novel jelly sh-inspired swimming robot whose mechanical design, control algorithm, and motion analysis are discussed and manifested in detail. In consideration of few research on robotic jelly sh capable of threedimensional movement with mechanical drive and actuator, we propose an improved design mainly featuring six-bar linkage mechanisms as the actuators and a barycenter adjustment mechanism as an attitude regulator. To imitate the real jelly sh and to reduce drag, the robotic jelly sh is comprised of a streamlined head, a cavity shell, and an elastic rubber skin around the drive units. Meanwhile a bio-inspired central pattern generators control model is imported. To validate the feasibility of the maneuverability with designed motion mechanism and free attitude regulation, dynamic analysis and simulation experiment are conducted. The obtained results show the conceived novel robotic jelly sh can accomplish all motion tasks desired.
In this paper, we develop a novel approximate policy iteration reinforcement learning algorithm with unsupervised feature learning based on manifold regularization. The proposed algorithm can automatically learn data-...
In this paper, we develop a novel approximate policy iteration reinforcement learning algorithm with unsupervised feature learning based on manifold regularization. The proposed algorithm can automatically learn data-driven smooth basis representations for value function approximation, which can preserve the intrinsic geometry of the state space of Markov decision processes. Moreover, it can provide a direct basis extension for new samples in both policy learning and policy control processes. We evaluate the effectiveness and efficiency of the proposed algorithm on the inverted pendulum task. Simulation results show that this algorithm can learn smooth basis representations and excellent control policies.
In this paper, an sEMG-driven musculoskeletal model of human shoulder and elbow joints is built based on time delay neural network (TDNN). Six principal muscles of the upper arm and forearm are included, and the exper...
详细信息
In this paper, an sEMG-driven musculoskeletal model of human shoulder and elbow joints is built based on time delay neural network (TDNN). Six principal muscles of the upper arm and forearm are included, and the experiment was conducted under isometric contractions with the aid of a planar haptic interface. Both force amplitude and direction were regulated continuously, and the experiment results proved the effectiveness and performance of this modeling method. The model was proved to have less overfitting risk than the most-used basic multilayer forward networks, and the isometric model was proved to be still effective in estimation of slow movement cases.
暂无评论