Open-Loop Motion Control of a Hydraulic Soft Robotic Arm Using Deep Reinforcement Learning

2021 ◽  
pp. 302-312
Author(s):  
Yunce Zhang ◽  
Tao Wang ◽  
Ning Tan ◽  
Shiqiang Zhu
2011 ◽  
Vol 121-126 ◽  
pp. 4613-4618
Author(s):  
Zong Min Chen ◽  
San Nan Yuan

This paper presents a Field Programmable Gate Array based closed-loop motion control system for stepper motors. It consists of three components, including closed-loop control unit, driving unit and feedback unit. To overcome some of the drawbacks with an open-loop stepper motor motion control system or a conventional servo system, a self adaptive algorithm is proposed. By detecting the difference between the command and feedback signals, measures are taken prior to the occurrence of loss synchronization. All of the control logic is implemented in one FPGA chip. Simulation and testing results are presented at the end of this paper.


Author(s):  
Zhi Qiao ◽  
Pham H. Nguyen ◽  
Panagiotis Polygerinos ◽  
Wenlong Zhang

2018 ◽  
Vol 18 (07) ◽  
pp. 1840017 ◽  
Author(s):  
QIN YAO ◽  
XUMING ZHANG

Flexible needle has been widely used in the therapy delivery because it can advance along the curved lines to avoid the obstacles like important organs and bones. However, most control algorithms for the flexible needle are still limited to address its motion along a set of arcs in the two-dimensional (2D) plane. To resolve this problem, this paper has proposed an improved duty-cycled spinning based three-dimensional (3D) motion control approach to ensure that the beveled-tip flexible needle can track a desired trajectory to reach the target within the tissue. Compared with the existing open-loop duty-cycled spinning method which is limited to tracking 2D trajectory comprised of few arcs, the proposed closed-loop control method can be used for tracking any 3D trajectory comprised of numerous arcs. Distinctively, the proposed method is independent of the tissue parameters and robust to such disturbances as tissue deformation. In the trajectory tracking simulation, the designed controller is tested on the helical trajectory, the trajectory generated by rapidly-exploring random tree (RRT) algorithm and the helical trajectory. The simulation results show that the mean tracking error and the target error are less than 0.02[Formula: see text]mm for the former two kinds of trajectories. In the case of tracking the helical trajectory, the mean tracking error target error is less than 0.5[Formula: see text]mm and 1.5[Formula: see text]mm, respectively. The simulation results prove the effectiveness of the proposed method.


2018 ◽  
Vol 31 (5) ◽  
pp. 608-622 ◽  
Author(s):  
Zhe Chen ◽  
Xueya Liang ◽  
Tonghao Wu ◽  
Tenghao Yin ◽  
Yuhai Xiang ◽  
...  
Keyword(s):  

2018 ◽  
Vol 8 (1) ◽  
Author(s):  
Brennan T. Phillips ◽  
Kaitlyn P. Becker ◽  
Shunichi Kurumaya ◽  
Kevin C. Galloway ◽  
Griffin Whittredge ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document