Design and characterization of soft continuum manipulators using fiber reinforced actuators

Soft Robotics;Continuum Arms

Uppalapati, Naveen Kumar

关键词: Soft Robotics;    Continuum Arms;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/107857/UPPALAPATI-DISSERTATION-2020.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Soft robots attain their mobility through the deformation of stretchable skins, fibers, tendons, and pressurized fluids. Their recent popularity is a result of their attributes, such as adaptability, safe human interaction, cost-effectiveness in manufacturing, and deployment. As a result, they find use in manipulation, locomotion, and wearable devices. This dissertation deals with the design, modeling, and control of soft continuum arms (SCAs) and end effectors used for manipulation, with applications in agricultural berry harvesting.In general, soft robotic manipulators suffer significant performance trade-offs. The quest to increase dexterity leads to increasing design complexity and inertia. While adaptable and safe, the soft and compliant nature of these robots leads to lower accuracy, precision, load-bearing ability, and speed. This dissertation aims to maximally overcome these trade-offs through systematic design and data-driven modeling.This dissertation utilizes pneumatic fiber-reinforced actuators as the basic building block for designing soft robots. An insight into the design of SCAs by exploiting asymmetry in the fiber angles (local scale) and by combining different fiber-reinforced actuator building blocks in an asymmetric parallel architecture (global scale) is presented. Asymmetry in the fiber angles was utilized to design and demonstrate a spiral gripper to grasp long and slender objects. Similarly, combining asymmetric building blocks has resulted in the design of a unique soft manipulator known as the BR2, with large spatial workspace and dexterity. Kirchhoff's rod model was used to capture the large nonlinear continuum elastic deformation of the SCA with a novel parameter estimation scheme to evaluate its actuation and elastic coefficients. The rod model is further used to predict the deformation of the soft manipulator under different loading conditions.Robust control of the asymmetric SCAs is challenging because it often requires intricate analytical and numerical models. The complexity of these models may render traditional model-based control difficult and unsuitable. In this dissertation, a model-free approach for position control of SCA, based on deep reinforcement learning is described.The design principles are used to create a hybrid robotic system from a combination of soft continuum arm and a rigid robotic arm. This system has the advantage of high payloads, compact architecture, and force transfer capability while simultaneously attaining large spatial workspace and dexterity. While the hybrid robotic system can have multiple applications, the design, system integration, control, and sensing is targeted towards an autonomous agricultural robot for harvesting berries.