Hua Deng

Adaptive Fuzzy PID Force Control for a Prosthetic Hand

An adaptive fuzzy proportional-integral-derivative (PID) force control strategy for a prosthetic hand is presented. The classical PID controller is also applied on the prosthetic hand as comparison. The parameters of PID controller are firstly tuned by Cut and Try method. Then a fuzzy logic system is used to adjust those parameters on line. Real-time force control experiments are realized on LabVIEW and PXI (PCI eXtensions for Instrumentation) real-time (RT) platforms. A rigid object and a compliant object are grasped by the prosthesis respectively to test the performance of controllers. Experimental results indicate that the adaptive fuzzy PID force controller is more effective than PID controller.

Dynamic Modeling and Grasp Characteristics Analysis of an Underactuated Prosthetic Finger

A dynamic model was established using the virtual spring approach for the underactuated prosthetic finger containing the closed kinematic chains and holonomic constraints. The dynamic model was verified by grasp simulation. The virtual spring is used to approximate the constraint force and differential algebraic equations are converted into ordinary differential equations which are ideal for simulation and real-time control. The grasp characteristics of the underactuated finger were studied based on the model with the stiffness of the linear springs as variables. By properly increasing the stiffness of springs, the grasp stability of underactuated finger could be improved and ejection phenomenon would be reduced. However, self-adaptive capability would be reduced with large stiffness. A characteristic index was used for estimating grasp stability and self-adaptive capability. The simulation results show that setting the stiffness of the linear springs between 1N/mm and 2N/mm is the best choice for the underactuated finger.

Model Reduction of Rigid-Flexible Manipulators with Experimental Validation

An investigation into the dynamic modeling of rigid-flexible manipulator systems based on spectral approximation methods was presented. The nonlinear dynamic equation of the rigid-flexible manipulator was established on the theory of Lagrange equation and the model was approximated with the spectral approximation method. Furthermore, an experimental platform system was set up to validate the effectiveness of the model. Simulation results of the response of the rigid-flexible manipulators were presented. Moreover, experimental validations were carried out to assess the effectiveness of the modeling approach by comparing with the simulation results. The results verified that the method is effective and feasible.

Grasping force control preventing excessive deformation for underactuated prosthetic hands

An anti-ejection control strategy was proposed for the characteristics of underactuated prosthetic hand grasping process. The reference position can be calculated and the stiffness of objects can be estimated by the feedback signals of force and position. When the driving joint approaches the reference position and the actual grasp force does not reach the desired value, which means that the ejection tendency of the prosthetic hand and the expected force need to be adjusted to attain the stable grasping status.

Gait Generation of Humanoid Robot Based on Analysis of the Human's Gait

In this paper, a method to generate the humanoid robots gait intelligently is put forward to solve the problem of poor performance of robot walking. The key idea in this proposed method is to adapt the transmutative humans gait to robot walking. Firstly, the character of the humans gait is acquired by researching a mass of gait data. Then, the typical gait signal is obtained which can be used to generate various gait signals. Finally, this method is proved to be effective by comparing the nature signals and the signal which is obtained by this method.

Optimal design for heavy forging robot grippers

This paper analyzes three typical mechanisms of heavy forging robot grippers: pulling with a sliding block including short- and long-leveraged grippers and pushing leveraged grippers, and uses multi-objective evolutionary genetic algorithm to design the optimal forging robot grippers. The decision variables are defined according to the geometrical dimensions of the heavy grippers, and four objective functions are defined according to gripping forces and force transmission relationships between the joints, and the constraints are yielded by the physical conditions and the structure of the grippers. Elitist Non-dominated Sorting Genetic Algorithm (NSGA-II) is used to solve the optimization problem. Normalized weighting objective functions are used to select the best optimal solution from Pareto optimal fronts. The Pareto fronts and optimal results are compared and analyzed. An optimal model of forging robot gripper is designed. The results show the effectiveness of the optimal design. Based on similarity theory, optimum dimensions from small scale forging grippers to large scale ones can be designed, and from model to prototype experiment to test the physical features is possible.

Hybrid Slip Detection Scheme Based Reflex Control of a Prosthetic Hand

Through the judgment of slip or not, human make proper adjustment to the grasping force and achieve stable manipulations. To reconstruct this function on a prosthetic hands platform, this paper presents a hybrid slip detect algorithm utilizing the PVDF sensor and FSR sensor. Then a reflex force estimation model is built to quantify the reflex force according to the intensity of slip in the reflex control process. Finally, through comparative experiments, the anti-jamming performance of the hybrid slip detect scheme is tested. A fuzzy controller is used to control the applied force and test the whole reflex control system. The results show that the hybrid slip detect scheme can make accurate judgment and has strong anti-jamming capacity; The output of the reflex force estimation model is accordance with the factual case; And as a whole, the grasping ability of prosthetic hand is substantially enhanced.

Spectral based spatio-temporal modeling for thermal crown of aluminium alloy hot rolling processes

A spectral based low-dimensional spatio-temporal modeling approach is proposed for thermal crown of work rolling in aluminium alloy rolling processes. Firstly，the Karhunen-Loève (KL) decomposition is used for dimension reduction and time/space separation. The neural networks are used for dynamic modeling. The simulations have demonstrated the effectiveness of the proposed spatio-temporal modeling approach..

The Research on Characteristics of Movement Coupling in Forging Manipulator

Heavy-duty forging manipulators play an important role in extreme manufacturing. Based on the kinematic model of DDS forcing manipulator, the input- output relation equations in terms of position are derived. According to the actually movement path of output mechanism, a new decoupling concept is defined as to forcing manipulator. The input-output relation matrices in terms of velocity are established at three major motions, and the coupling characteristics of mechanism are analyzed. At last, the theory analysis is proved correct by experimental verification. The results show that three major motion mechanisms of DDS forcing manipulator are partially decoupled under the new decoupling concept compared with the conventional one, which helps to simplify the control strategy of DDS forcing manipulator.

Model Reduction of a Two-Link Rigid-Flexible Manipulator Based on Spectral Approximation Method

In this paper, the dynamic model of a two-link rigid-flexible manipulator was established by using the theory of Lagrange equation. Due to the model is a distributed-parameter system, we present a simple, efficient method for model reduction based on spectral approximation method such that a model-based scheme can be implemented. The method was illustrated by comparing with the simulation results of the reduced-order model and the virtual reduced model based on finite element method.