Search

Keypoints matching is a pivotal component for many image-relevant applications such as image stitching, visual simultaneous localization and mapping (SLAM), and so on. Both handcrafted-based and recently emerged deep learning-based keypoints matching methods merely rely on keypoints and local features, while losing sight of other available sensors such as inertial measurement unit (IMU) in the above applications. In this paper, we demonstrate that the motion estimation from IMU integration can be used to exploit the spatial distribution prior of keypoints between images. To this end, a probabilistic perspective of attention formulation is proposed to integrate the spatial distribution prior into the attentional graph neural network naturally. With the assistance of spatial distribution prior, the effort of the network for modeling the hidden features can be reduced. Furthermore, we present a projection loss for the proposed keypoints matching network, which gives a smooth edge between matching and un-matching keypoints. Image matching experiments on visual SLAM datasets indicate the effectiveness and efficiency of the presented method.

To satisfy the requirements of the Plan for Education and Training Outstanding Engineers, teaching reform group has implemented a reform of the digital circuit curriculum. The reform is based on the problems of the traditional digital circuit course. The new technology of FPGA is introduced and the curriculum knowledge structure is adjustedensuring the integrity and comprehensiveness of the knowledge system. Moreover, the novel teaching methods is applied to practice with a development kit (pocket lab). Besides, novel assessment methods are proposed to nurture students’ innovation and application abilities. After two rounds of experiments, the students’ comprehensive ability has been strengthened, and the teaching effect of the reform is obvious.

This paper introduces the structural design of a novel spherical motion generator. The electromagnetic drive mode of a spherical motor is applied to the three-degree-of-freedom (3-DOF) spherical parallel manipulator, in order to solve the problem of motion error caused by the transmission device and motor following in the traditional spherical parallel manipulator. A brief explanation of the electromagnetic drive working principle is mentioned. And the modeling of magnetic field and the optimization of magnetic field distribution are also discussed. In the future work, we will also optimize the magnetic field distribution of the platform and study the high-precision control strategy.

Nonuniform web tension has a great negative effect on product quality of roll-to-roll printed electronics. In this paper, a roller posture adjustment device with remote-center-of-motion (RCM) characteristic is proposed to guarantee web tension uniformity. The device combines a high-stiffness spherical air bearing (SAB) and a multi-degree-of-freedom flexure-based mechanism. The nonuniform web tension in lateral direction can be divided into an equivalent force and two equivalent moments. The equivalent moments are caused by the nonuniformity of the web tension. By adjusting the roller angle around y and z-axes, the equivalent moments can be eliminated to guarantee the web tension uniformity. The multi-DOF flexure-based mechanism is composed of a linear mechanism, a rotary mechanism and a 3-DOF off-plane mechanism. Besides, the RCM characteristic of the proposed device is realized to eliminate extra parasitic movement when the roller posture is changed. Based on pseudo-rigid-body model (PRBM) method, the theoretical analysis is conducted to evaluate the kinematics, stiffness and dynamics of the device. Moreover, the parameter optimization is conducted to maximize the first two resonance frequencies of the system. After that, finite element analysis is conducted to validate the established models. Finally, a prototype of the proposed device is fabricated performance verification. The experimental results show that the proposed device has a workspace of 10.22 mrad and 8.16 mrad about two working axes with center shifts of the RCM point less than 0.75%, which demonstrate the superior property of the proposed device.

In this paper, we present the prototype of a novel spherical motion generator driven by an electromagnetic force. It is such a mechanism that combines the structure of a spherical parallel manipulator with the actuation of the spherical motor. To accomplish the task of trajectory tracking, we firstly employ the Lagrange-D’Alembert equation to establish the dynamic model. Due to the nonlinearity of the system, we attempt to apply three different control methods to compare their performances. Simulation model is formulated by Matlab/Simulink and the experimental results demonstrate that all control strategies can almost achieve the goal of trajectory tracking. The computed- torque method (CTM) has a better trajectory tracking performance than the others in the presence of friction for that it makes use of the dynamic information and converts a nonlinear system into a linear one via full-state nonlinear feedback.

A permanent magnet spherical actuator is an electromagnetic device characterized by achieving multi-degree-of-freedom rotations in one driving unit. Analytical magnetic field modeling of the rotor array is needed for the design, performance analysis and optimization of the spherical actuator. This paper presents an analytical method to formulate the complex magnetic field distribution of rotor array with cylindrical permanent magnets in 3D spherical space. This method combines the current model by toroidal functions and transformation method, thus the magnetic field can be calculated efficiently with reasonable accuracy in order for further design optimization and real-time control. The analytical magnetic model of the permanent magnet rotor array is validated with the numerical finite element method by Ansoft Maxwell and experimental measurement. A permanent magnet spherical actuator is an electromagnetic device characterized by achieving multi-degree-of-freedom rotations in one driving unit. Analytical magnetic field modeling of the rotor array is needed for the design, performance analysis and optimization of the spherical actuator. This paper presents an analytical method to formulate the complex magnetic field distribution of rotor array with cylindrical permanent magnets in 3D spherical space. This method combines the current model by toroidal functions and transformation method, thus the magnetic field can be calculated efficiently with reasonable accuracy in order for further design optimization and real-time control. The analytical magnetic model of the permanent magnet rotor array is validated with the numerical finite element method by Ansoft Maxwell and experimental measurement.

This paper presents modeling of an integrated design of an electromagnetic driven spherical motion generator with multi-layer structure. An analytical method to calculate the complex magnetic field distribution of the rotor array by combining the equivalent charge model and transformation method is developed, upon which torque model is further established analytically by using the moment principle. Both the analytical magnetic model and torque model are validated with the numerical finite element method by Ansoft Maxwell and experimental measurements. With the model, two different designs of the electromagnetic driven SMGs are analyzed to illustrate the application of the developed model. This paper presents modeling of an integrated design of an electromagnetic driven spherical motion generator with multi-layer structure. An analytical method to calculate the complex magnetic field distribution of the rotor array by combining the equivalent charge model and transformation method is developed, upon which torque model is further established analytically by using the moment principle. Both the analytical magnetic model and torque model are validated with the numerical finite element method by Ansoft Maxwell and experimental measurements. With the model, two different designs of the electromagnetic driven SMGs are analyzed to illustrate the application of the developed model.

The applications of robotic fish require high propulsive efficiency mechanism to prolong the mission time. Though many methods were applied, robotic fish still suffers from low efficiency. To improve the efficiency of robotic fish, this paper proposes a variable stiffness mechanism which is based on the negative work. The live fish adjusts its body stiffness to save energy when the muscles do negative work. Inspired by the live fish, a control mechanism based on negative work is proposed to change the stiffness of the robotic fish for higher efficiency. Changing the stiffness of the robotic fish is to change the joint-stiffness. A fuzzy controller is introduced to mimic the variable stiffness mechanism of the fish and depicts the relationship between the stiffness and the negative work. To evaluate the performance of this controller, a two-joint robotic fish model is established based on its kinematic model and hydrodynamic model. The evaluation results show that the robotic fish reduces the energy consumption and improves the propulsion efficiency when introducing the variable stiffness mechanism. Different environments with the control mechanism impact differently on propulsive efficiency. This mechanism may provide a high efficient propulsion control method for the robotic fish.

In this paper, a robust adaptive control system combining backstepping and sliding mode control method has been implemented for the 3-DOF permanent magnet (PM) spherical actuator in order to improve its trajectory tracking performance. Due to the complexity of both mechanical structure and electromagnetic field, the dynamic model of a PM spherical actuator inevitably contains uncertainties, such as disturbances and model errors, which will gravely influence the performance of conventional control system. Therefore, a backstepping sliding mode approach is firstly applied, where the backstepping design is to synthesize the controller and the sliding mode term is to compensate disturbances. Then, an adaptive law is presented to estimate the model errors during the control process, where the parameters of the model are initially assumed to be accuracy. Correspondingly, the stability can be guaranteed by choosing the appropriate Lyapunov function. Both simulations and experiments are designed to demonstrate the effectiveness of the proposed control scheme.

Short-term traffic forecast is one of the essential issues in intelligent transportation system. Accurate forecast result enables commuters make appropriate travel modes, travel routes, and departure time, which is meaningful in traffic management. To promote the forecast accuracy, a feasible way is to develop a more effective approach for traffic data analysis. The availability of abundant traffic data and computation power emerge in recent years, which motivates us to improve the accuracy of short-term traffic forecast via deep learning approaches. A novel traffic forecast model based on long short-term memory (LSTM) network is proposed. Different from conventional forecast models, the proposed LSTM network considers temporal-spatial correlation in traffic system via a two-dimensional network which is composed of many memory units. A comparison with other representative forecast models validates that the proposed LSTM network can achieve a better performance.

In order to avoid the damage of collision between the assembly targets during microassembly, a new 5-degree- of-freedom (5-DOF) fleuxre-based passive alignment stage for microassembly system is presented. With fiber assembly as the application field, the design, modeling and simulation of the proposed stage are proposed. The proposed stage mainly consists of a 3-DOF in-plane mechanism and a 2-DOF off-plane mechanism. The alignment motion of the stage is achieved by adopting a variety of flexure hinges. Due to the remote center of compliance (RCC) characteristic, the proposed stage can compensate position error in different directions decoupleably. In order to guarantee successful fiber assembly process, the stiffness of the proposed stage is strictly controlled. The pseudorigid-body-model (PRBM) method is utilized to calculate the stiffness of the stage along its working directions. Then the finite element analysis (FEA) is conducted to verify the stiffness theoretical model and test the dynamic performance of the stage. Both the theoretical and FEA results demonstrate the superior property of the design, which indicates that the proposed stage can meet the requirements for compensating the relative position error between the fiber and the optical switch.

This paper investigated the problem of moving target recognition and tracking in unknown environments for autonomous mobile robots. The mobile robot is equipped with the Kinect sensor to acquire the appearance and depth information of the environment for target recognition and motion control. As for the visual target recognition, the depth feature is fused with the color feature on the basis of the Kernel Correlation Filtering algorithm (KCF), and the Kalman filtering algorithm based on the target motion state is integrated with the KCF, to cope with such situations as the target occlusion in tracking process. The movement of the robot can be controlled according to the results of visual target recognition and the standard of keeping the target in the central area of the mobile robot’s vision. Finally, controlling the mobile robot to track the moving target is realized. The experimental results on Kobuki robot platform prove the effectiveness of the presented method.

The development of electronic and computer technology makes non-invasive acquisition systems of EEG more universal adoption. Therefore this paper focus on the BCI based on motor imagery, which is a popular studied spontaneous EEG, producing without external stimulus. 6 healthy people participate in the examination, 5 of which are saved while one of which is thrown due to disturbing. Independent component correlation algorithm (ICA) was used to extract motor related component, meanwhile other filters also designed for better signal quality. Taking the motor related components PSD distribution after ICA as features, signal modal decoding based on reinforcement learning is processed in two algorithms. Both of them have good performance. Finally, on-line decoding model is descripted on the basis of the off-line decoding model. The result shows the accuracy of training model has significant difference with random level. Furthermore, the conclusion can be developed into human-computer integration based on EEG signals.

This study presents a robust augmented nonlinear proportional derivative control (ANPD) strategy to solve the trajectory tracking problem for the spherical motion generator (SMG). The dynamic model of SMG obtained by Lagrange-Euler formalism is obviously a multi-variable nonlinear system with external disturbances, which will inevitably affect the precision of the control system. In the ANPD controller's framework, there exist three compensation items comprised of dynamics compensation term, disturbance compensation term and nonlinear error elimination term. The ANPD control algorithm aims at offsetting the effects of external perturbations and realizing accurate control. To illustrate the effectiveness of the proposed control algorithm, simulation model is formulated in Matlab/Slimulink. The results show that the proposed control algorithm has better competence of trajectory tracking in the presence of uncertainties.

The dynamic model of multi-degree-of-freedom permanent magnet (PM) spherical actuators is multivariate and nonlinear due to strong inter-axis couplings, which affects the trajectory tracking performance of the system. In this paper, a decentralised control strategy based on adaptive fuzzy sliding mode (AFSM) algorithm is developed for a PM spherical actuator to enhance its trajectory tracking performance. In this algorithm, the coupling terms are separated as subsystems from the entire system. The AFSM algorithm is applied in such a way that the fuzzy logic systems are used to approximate the subsystem with uncertainties. A sliding mode term is introduced to compensate for the effect of coupling terms and fuzzy approximation error. The stability of the proposed method is guaranteed by choosing the appropriate Lyapunov function. Both simulation and experimental results show that the proposed control algorithm can effectively handle various uncertainties and inter-axis couplings, and improve the trajectory tracking precision of the spherical actuator. The dynamic model of multi-degree-of-freedom permanent magnet (PM) spherical actuators is multivariate and nonlinear due to strong inter-axis couplings, which affects the trajectory tracking performance of the system. In this paper, a decentralised control strategy based on adaptive fuzzy sliding mode (AFSM) algorithm is developed for a PM spherical actuator to enhance its trajectory tracking performance. In this algorithm, the coupling terms are separated as subsystems from the entire system. The AFSM algorithm is applied in such a way that the fuzzy logic systems are used to approximate the subsystem with uncertainties. A sliding mode term is introduced to compensate for the effect of coupling terms and fuzzy approximation error. The stability of the proposed method is guaranteed by choosing the appropriate Lyapunov function. Both simulation and experimental results show that the proposed control algorithm can effectively handle various uncertainties and inter-axis couplings, and improve the trajectory tracking precision of the spherical actuator.