Design And Control Of A Tendon-Driven Soft Robot Manipulator

With the development of science and technology,the traditional rigid and discrete robots are becoming more and more popular in the industrial field.Rigid robots perform well in structured and non-constrained environments,but are hard to meet the demand of dynamic and unstructured environments.Soft robots,which mimic the kinematic mechanism of molluscs in nature,are very suitable for unstructured environments because of their infinite degree of freedom,high flexibility and high safety.And soft robots have great prospects in the field of medical services,disaster relief,scientific exploration and military reconnaissance.The goals of this paper were as following: to design a soft robot manipulator and optimize the structure and parameters of the robot through experiments,to study and optimize its control model and improve kinematic accuracy of the soft robot manipulator,to explore potential applications of the soft robot manipulator.Through the comparison with other drive methods,tendon-driven was chosen to control the soft robot manipulator.Based on the piecewise constant curvature model,the kinematic analysis was carried out to study the mapping relationship among three spaces: actuator space,configuration space and task space.The workspace and kinematic property of the soft robot manipulator were simulated by MATLAB.An experimental prototype platform was developed and constructed,including an upper computer control software,motion control system and mechanical bench.According to the various factors that affect the performance of the soft robot manipulator,such as specifications of the manipulator and some external conditions,several sets of contrast experiments were done.The structure and parameters of the soft robot manipulator were optimized according to the experimental results.Based on the discrepancy between experimental value and theoretical value,the control model was compensated to some extent,including control compensation under single variable and under multiple variables.Curve fitting and BP neural network were respectively adopted.The effectiveness of control model compensation was verified by experiment,and the control model of single-section soft robot manipulator was extended to multi-section soft robot manipulator.Some experiments of multi-section soft robot manipulator in potential application scene were carried out.