Data-based Inverse Kinematic Control for Multi-section Soft Manipulator

• Victor Santos Matos SENAI CIMATEC
• Luis Victor C. Novaes SENAI CIMATEC
• João Vitor S. Mendes SENAI CIMATEC
• Lucas Cruz Da Silva SENAI CIMATEC

Resumo

Soft manipulators present safe interaction with facilities and forms of life, and inherent dexterity, that demands a complex control system to precisely operate its theoretically infinite degrees of freedom. This article presents a form of data-based inverse kinematic (IK) control for a soft tendon-driven manipulator, with a closed-loop form to help the disturbance reduction to deal with multiple actuated sections, where the position of the end-effector and the middle-section can be explicitly defined. The IK is mapped by a neural network based on simulation training data from the SOFA framework simulator, with the evaluation for the best structure. Tests were conducted in the simulated manipulator, indicating a satisfactory control performance, even in an unstructured environment.

Palavras-chave: inverse kinematics, soft manipulator, data-based, neural network, multi-section control, simulation