Towards Dexterous Agri-Food Manipulation: Topology-Dependent Interaction Patterns in a Reconfigurable Multifingered Gripper

Hongyu Lan, Alessio Caporali, Chengxiao Dong, Gianluca Palli, Claudio Melchiorri

Keywords: Gripper, Grasping, Agricultural Automation

Abstract: Multi-fingered and reconfigurable grippers offer greater adaptability than fixed designs, but it remains unclear how grasp topology systematically affects object-level interaction patterns in contact-rich manipulation. In this paper, we present a contact-rich simulation framework in AGX Dynamics to analyze topology-dependent grasp behavior of a reconfigurable four-finger gripper under controlled yaw and planar-offset perturbations across representative agri-food objects. Beyond grasp success, we evaluate mechanics-level interaction patterns to compare robustness, torque response, friction demand, load redistribution, and penetration-based deformation tendency. The results show that more enveloping topologies achieve consistently higher robustness to planar placement uncertainty, while object-centered torque is markedly more sensitive to perturbations than contact force and thus provides a more discriminative indicator of topology-dependent robustness. We further find that friction demand is governed primarily by object geometry, whereas grasp topology more strongly affects how interaction mechanics respond to perturbations. These findings provide quantitative guidance for grasp configuration selection and for evaluating reconfigurable multifingered grasping systems beyond success rate alone.

Submission Number: 15