Performance Improvement of a Class of Continuum Manipulators via Adaptive Algorithms

Abstract

Continuum manipulators represent a class of robots reproducing some bio-inspired behaviors such as elephant trunks, octopus, or tentacles. Their performances in terms of speed limitation and position accuracy are often mediocre compared to rigid body-based robots. The presence of nonlinearities effects and the hyper-redundancies in continuum manipulator structure makes it difficult to establish analytical accurate kinematic models that can be integrated into a real-time control scheme. Unlike in rigid body-based manipulators where accurate kinematic models generally lead to good performance in closed loop, in the case of continuum manipulators, accurate kinematic models do not always give satisfactory results. The structure of some continuum manipulators would be the main cause of this poor performance. Here, continuum manipulator performance is measured in terms of positioning accuracy. The paper attempts to demonstrate that even using global optimization learning methods which guarantee global solutions, it remains a challenging task to achieve good performance with non-adaptive control laws. This is demonstrated by implementing adaptive and non-adaptive algorithms on a class of continuum manipulators namely the compact bionic handling arm (CBHA).