Design, Modeling and Motion Control of a Multi-Segment SMA Driven Soft Robotic Manipulator

Abstract

Soft Roboic Manipulators (SRMs) have shown potential for their ability in complex motions. In this work, a biomimetic soft robotic manipulator driven by Shape Memory Alloy (SMA) is developed. The kinematic model and sensor system of the SRM are built. The rotation matrix and the Jacobian of the SRM are presented for the solution of the forward kinematics. The visual perception based on deep learning is used to perceive the objects and the angular sensors are used to obtain the attitude information of the SRM. Each set of SMA actuators can be actuated independently to allow the manipulator to generate a variety of complex motions. Due to the special structure designed, the SRM can drive its 5kg body to complete many complex movements with only a few combinations of SMA. A control method based on attitude perception for the SRM is developed, which can output the control sequence of SMA directly. Experiments show that the SRM can perform motions such as object grasping and obstacle avoidance tasks. The angular velocity and velocity during the a object grasping task are analyzed. The experiments indicate the SRM’s great potential in complex applications.