Kinematic Modeling and Self-Sensing of a Shape Memory Alloy Wire-Driven Continuum Robot

Due to their unique manipulation capabilities and flexibility, continuum robots find applications in different areas such as industry, medicine, and security. Shape Memory Alloy (SMA) wire transducers represent one of the most suitable technology to drive continuum robots, due to their high energy density, flexibility, and ability to work as actuators and sensors at the same time. More specifically, the length of an SMA wire can be reconstructed during actuation based on the information provided by its electrical resistance. Resistive self-sensing can be used to estimate in real-time the overall configuration of the robot without additional electro-mechanical sensors or cameras, thus reducing the overall system size and complexity. Clearly, implementation of self-sensing based motion control strategies still requires the availability of mathematical models for SMA-driven continuum robots. This paper presents a twofold experimental contribution in modeling and self-sensing for SMA-driven continuum robots. First, the robot electro-mechanical design is discussed, and the experimental setup used for data acquisition and control is presented. The measured motion data are then used to validate a kinematic model of the device. Finally, a neural network is trained based on the acquired data, showing the feasibility of reconstructing the robot motion in 3D using only actuators resistance information. The presented results will enable future implementation of self-sensing based motion control systems for SMA-driven continuum robots.