The Design of a Vision-Based Bending Sensor for PneuNet Actuators Leveraging ArUco Marker Detection

Conventional bending sensors mainly utilize electrical, chemical, or optical signals to control the bending action of the soft pneumatic bending actuator. Recently, image processing techniques have emerged as an alternative for detecting such bending movements. However, their fabrication process and measurement methodology still remain intricate. In this article, a novel vision-based bending sensor for the pneumatic network (PneuNet) actuator based on ArUco marker detection is proposed. The fabrication of the proposed sensor is simplified through the use of 3-D-printed or outsourced parts, making the proposed sensor’s design and fabrication much easier. The PneuNet is connected to a rotating disk featuring an ArUco marker via an inextensible Kevlar cable. Thus, the bending of the PneuNet is converted to the one degree of freedom (DoF) rotation of the ArUco marker, which can then be monitored by the camera module. Experiments show that a linear correlation between the bending and ArUco marker angles is characterized despite the hysteresis of the PneuNet, proving the proposed sensor’s usability. Moreover, control of multiple actuators is feasible as long as the markers are within the field of view (FOV) of the camera module. Finally, based on the mechanical detection method of the sensor, tracking of the PneuNet’s tip displacement upon the exertion of external force is also showcased to utilize the proposed sensor’s versatility further.