Research On Design,Control And Application Of Soft Robotic Arm Based On Honeycomb Pneumatic Networks Structure

In order to better liberate humans from heavy and dangerous labor,robots need to have the ability to manipulate in unstructured environments,which is a challenge in the field of robotics.Soft robotic arms bring new possibilities for solving the manipulation problems in unstructured environments thanks to their passive compliance and continuous deformation.However,due to the limitations of the design and control of the soft arm,no one has yet explored the possibility of the soft arm performing manipulation tasks in an unstructured environment.In order to explore this possibility,the thesis first explores the design of the soft arm.Based on the honeycomb pneumatic network(HPN)structure,a soft arm with both flexibility and load capacity is designed.Secondly,the thesis explores the control problem of the soft arm and proposes three new control methods:The first method combines the advantages of learning-based modeling and modeling based on simplified assumptions to establish a model,based on which the open-loop control of the soft arm is realized.The second method establishes a simplified differential kinematics model based on the motion law of the soft arm that is not affected by the environment,and then uses incremental PID adjustment to realize the closed-loop control of the soft arm.The third method takes the soft arm and the environment it interacts with as a whole,and uses reinforcement learning to achieve its closed-loop control,in which we proposes a novel method to increase training data by setting virtual goals.After solving the design and control problems,the thesis,inspired by the way humans perform interaction tasks,proposes a hierarchical control system for the soft arm.Through a series of experiments on daily manipulation tasks,it is verified that the soft arm can simply complete interactive tasks without the need for force sensors or accurate modeling of the environment.Finally,with the problem of feeding assistance for disabled people as the background,the paper demonstrates the advantages and application potential of soft arms in solving practical manipulation problems.This work provides a new direction for the application of soft arms and offers a new perspective for the physical interactions between robots and environments.