机器学习、人机交互

模型预测控制、神经网络

多智能体系统、网络控制系统等等

2020年最后一期

来看你想要的！

• Toward understanding nonverbal cues and signals in human-robot social interaction.

• A face-to-face human robot interaction from the robot's first-person perspective.

• A new computational framework for enabling social robots to understand human personality traits.

• Performance evaluation of different nonverbal social cues using machine learning techniques.

• This paper proposed a tracking controller for pneumatic muscle actuators driven exoskeleton.

• Based on the neural network approximation model, this controller is a data-driven strategy.

• The controller is proven to be asymptotically stable.

• Experimental results indicate the effectiveness and robustness of this controller.

• A data pre-processing method is raised before data fusion, which could solve the time alignment problem.

• To improve the accuracy of data fusion system, a data fusion approach based on reinforcement learning is designed by multi-sensors weight adjustment.

• In the case without prior knowledge, the reinforcement learning based data fusion is realized by the Fisher information of observations.

• A new problem of Group Role Assignment with Constraints (GRA+), called Group Multi-role Assignment (GMRA) with Conflicting roles and agents (GMAC) is proposed and formalized.

• A practical solution based on an optimization platform, i.e., IBM ILOG CPLEX Optimization Package is provided.

• A sufficient condition, used in planning, for solving GMAC problems is proved.

• The benefits of avoiding conflicts when dealing with GMAC are presented.

• Asynchronous clock, mobility and privacy preservation are together considered.

• Asynchronous localization protocol can effectively eliminate asynchronous clock.

• Asynchronous localization algorithm can effectively hide privacy information.

• Location accuracy can be guaranteed as compared with the others works.

• The core findings: We first found that multiple basic models and traditional algorithm description can be used to model various factors of user behavior, and finally integrated into a comprehensive user behavior model. 

• The essence of the research: We integrate a variety of user behavior factors, and also consider the characteristics of the external environment when using mobile devices. We synthesize several basic models to realize our method.

• The distinction of the paper: We have effectively integrated a variety of machine learning models and studied a series of algorithms on the models. We don't rely purely on machine learning models.

• Quick textual overview: We build a comprehensive model of user behavior, which integrates multiple user behavior factors with external environmental data.

• Our idea is to use different machine learning models to process heterogeneous multi-source data, and to build algorithms on top of machine learning models.

• The method in this paper realizes the iterative generation of feature vectors from the most primitive sensor data to realize the training of machine learning models. And real-time user behavior authentication using basic data is realized.

• Design a kind of vehicle tire longitudinal forces optimization distribution strategy, which increases the utilization rate of each wheel.

• Proposed a kind of compensation allocation strategy based on the axle load distribution method, which considers the tire longitudinal force constraint. And the effectiveness of actuators can be guaranteed.

• Adopt the hierarchical control structure for controller design, which effectively improves the performance of automotive active safety.

• The stabilization parametric region of distributed PID controllers for general first-order multi-agent systems with time delay is acquired in this paper, and the proposed method is conducted for systems under arbitrary topology including undirected and directed graph topology. Moreover, we ensure that the parameters chosen in the resultant set can guarantee the consensus of the multi-agent system with time delay.

• In the actual engineering application, the distributed PID controllers are generally required to meet several performance criteria simultaneously. For this phenomenon, the results in this paper solve the stabilization problem of the systems with complex coefficients, and provide the basis for the design of distributed PID controllers satisfying different performance criteria.

• The multi-agent system can be decoupled into several subsystems with respect to the eigenvalues of the Laplacian matrix based on the matrix theory. For each subsystem, the range of admissible proportional gains is analytically determined. Then, the stabilizing region in the space of integral gain and derivative gain for a given value is obtained in an analytical form. Finally, the entire stabilizing set can be determined by sweeping in the allowable range.

• Prior information from an accurate tracking SLAM is used to associate dense vertices between keyframes based on multithreaded processing and multi-threaded priority settings.

• The angle change and position change of the associated vertices are constructed and examined to determine if they are within two setting ranges to remove outliers. The two ranges are designed by using a rotation angle histogram and a beam-based environment measurement model, respectively.

• An adaptive weight is assigned to each inlier and the weighted fusion is implemented as the update process of the Kalman filter.

• The surfaces of inliers are stored in a global hash table and a local hash table for fast data operation and data reuse.

• Coordinate Conversion In order to obtain the accurate angular velocity and acceleration to control the speed of the electric bicycle, we calculate and use the real-time data of the carrier coordinate system (pedal) relative to the geographic coordinate system by quaternion-based coordinate conversion.

• Disturbance observer Considering the variability of frictions, we build a second-order low pass filter with a special cut-off frequency , so that the majority of disturbances can be observed and compensated.

• Velocity control A permanent magnet brushless DC motor is used for adaptively assisted riding. Based on the real-time angular velocity of the rear wheel, the motor can provide assistance power for controlling the riding speed within a certain range.

• When the smallest eigenvalues are equal, the dynamic characteristic analysis of M?ller algorithm is finished.

• Some conditions are drawn to guarantee the convergence of the Möller algorithm.

• The range of the learning rate is expanded.

• The influence of frequency drift on the separation of collision tag signals is considered.

• An adaptive radial basis function (RBF) neural network is introduced to separate the collision tag signals in a UHF RFID system.

• An FM0 decoding algorithm is proposed to decode separated tag signals by RBF.

• This paper constructs an observable measurement model under the BMU for absolute clock in the large-scale network from the viewpoint of networked control theory.

• Aiming at the calculation problem of positive definite invariant sets, the observability measurement model of MMSE-like equivalence is proposed to calculate the positive definite invariant set by using the Luenberger observer as the synthetical observer. Then the on-line calculation of the Tubes-MPC method for clock synchronization can be realized.

• Using the feedback control strategy and set-theory-in-control to establish the control error positive definite set, quantitatively analyzing the deviation between the estimated system state and the nominal system state with a measurement model of observability. The exponential stability convergence performance of Tubes-MPC for clock synchronization is achieved.

• The determined estimated value of the absolute state by using the equivalent observability measurement model of MMSE is based on the ideal state as public reference target. The problem of clock synchronization is transformed into the problem of set-point tracking to ideal state for local clock.

• In RACO-HDG, phase generation is designed for self-detector by using a small number of self-detector instead of a large number of anomaly detectors.

• In order to achieve optimal radius for each detector, self-adapt radius threshold is put forward based on self samples distribution.

• For reducing size of nonself-detector, nonself-detectors are generated from far to near.

• RACO-HDG reduces the number of self-detector and anomaly detectors to 1/10, and increases the true negative rate and reducing the false alarm rate.

• A new sliding mode variable is constructed, containing the message of DoS attack.

• A sliding mode control algorithm is developed under DoS attack.

• The developed sliding mode control algorithm can effectively deal with DoS attack.

• An important intrinsic natural feature of outdoor scenes free of sand/sandstorm is found.

• A tensor least square optimization model is presented for the decomposition of edge-preserving base layers and details.

• The tensor least square optimization model based image enhancement scheme is discussed for the sandstorm weather images.

• Optimal parallel tracking control for general nonlinear systems via ADP is studied in this research. Different from the optimal state feedback controller, the variation of the control of the optimal parallel controller is constructed based on both system state and control input.

• An augmented system and an augmented performance index function are proposed.

• It is proven that the optimal parallel control with the augmented performance index function can be seen as the suboptimal state feedback control with the traditional performance index function.

• The ADP method is utilized to implement the optimal parallel control without using value iteration or policy iteration.

• We prove that the tracking error state and the NN weights error are UUB and control input is continuous with the optimal parallel controller. Finally, the effectiveness of the developed optimal parallel control method is verified in two cases.