ReconfiGrip: A Single-Motor Reconfigurable Mechanism for Vision-Guided Soft Grasping

Reconfigurable robotic grippers can adapt to objects with different shapes and sizes. However, Most reconfigurable grippers rely on multiple actuators or complex mechanisms, which increase mechanical and control complexity. This paper presents a compact gripper driven by a single actuator that reliably switches between three discrete finger configurations using a cam-follower mechanism. The proposed design enables repeatable reconfiguration with one motor and minimal sensing, while keeping the mechanical structure simple and compact. Soft pneumatic fingers are used to handle local shape differences and minimize sensitivity to small positioning errors during grasping. A YOLO-based perception module identifies object type and orientation and selects the appropriate configuration before grasping. Experimental results show that adaptive operation increases grasp success from approximately 70% to 85% for artificial fruit items and from 66% to 84% for LEGO objects, compared to fixed finger configurations. These results demonstrate that discrete mechanical reconfiguration can improve grasp reliability without increasing actuation or control complexity.