Motion and force control in high-performance variable reluctance finger gripper

Abstract

Gripper mechanism is commonly found in industrial applications. An ideal gripper should have a simple and low-cost structure, fast and robust response. Various types of grippers have been designed but none of them is capable of fulfilling the requirements of an ideal gripper. Variable Reluctance (VR) actuator has a simple and robust structure, therefore it has been used in low-cost industrial applications. Together with its high-torque density in compared with classical DC motors, VR actuator can be a very attractive solution in gripping mechanism. However, due to its nonlinear characteristics, high-precision gripping applications tend to avoid it by employing other actuators which require less control effort. As a result, VR actuator is not popular in both industrial application and academic research. Due to the recent advancement of semiconductor components and micro-processors, more complex simulation model and advanced controller can be realized. In the past few years, VR actuators have regained much research attention. The project aims to investigate the feasibility of employing VR technology in precision two-finger gripping application. Under this overall goal, initial research efforts have been devoted to the analysis of VR actuator operating principles and the actual gripper design. Results have shown that torque density can be raised by 3 -4% for less than 2A and more than 30% with the use of mutual flux coupling effect when the pole-faces are saturated with 4A. This is achieved by connecting the flux return paths of two single VR fingers together. In order to fully understand the behaviour of the two-finger VR gripper prototype, detailed characterization experiments have been carried out. Flux-linkage and torque profiles are measured with Alternate Current (AC) current excitation method and direct torque measurement respectively. Measurement results confirm that the VR gripper prototype behaves like a VR actuator and enjoys an increase in efficiency once it operates within the saturation region. Other magnetic characteristics including leakage flux, hysteresis loss and eddy current loss, and spring stiffness have also been measured. After reviewing various modeling techniques commonly used in VR actuators, the nonlinear characteristics of the VR finger gripper, flux-linkage and torque profiles, are modeled with an exponential description function. Then, a concise dynamic nonlinear model of the VR finger gripper basing on state equations is established. This model is further verified by step responses and confirmed to be an accurate mathematical description for the VR finger gripper. An adaptive current regulation scheme is proposed and implemented. By continuously monitoring the position and current information, flux-linkage level can be calculated with the nonlinear model. Then control parameters are varied aiming at providing fast and repeatable current responses regardless of the rotor positions. Experimental results confirm that the proposed current regulation scheme is an effective solution for adapting to the varying inductance of the VR finger gripper. Two different control algorithms have been newly applied to tackle the nonlinearity of the VR finger gripper. The first method is to use a reduced-size nonlinear torque-current-position lookup table compensator. The main benefit of this method is simple to implement and straightforward. To further enhance the smoothness and accuracy of the table, a two dimensional interpolation scheme is proposed. Experimental results show that the proposed controller is capable of handling general applications. However, this control method fails to ensure the robustness and stability of the controlled system. In order to improve robustness qualities, a second method, the Passivity-based Control (PBC) is applied. This nonlinear control method bases on the concept of energy transformation within a system. Moreover, the controller design of this scheme is efficient and robust. Flux-linkage and trajectory control of the two-finger VR gripper using PBC are simulated with the mathematical model and implemented. Results show that the PBC controller is robust, offers better performance than the reduced order lookup table compensator and suitable for high-performance applications. Finally, the two-finger VR gripper is controlled under a mixed-mode control with command scheduling. Under such arrangement, the two-finger VR gripper can be precisely controlled under closed-loop position and force control. Results show that the controller's performance is highly satisfactory with good accuracy, high trajectory performance and fast dynamic response. Conclusively, the research work confirms that the proposed two-finger VR gripper is a appropriate solution for gripping mechanism.