Robotic gripper mimics human hand to move multiple objects together

A research team from Seoul National University has proposed a gripper capable of moving multiple objects together to enhance the efficiency of pick-and-place processes, inspired by humans’ multi-object grasping strategy. The gripper not only transfers multiple objects at once but also places individual objects at desired locations. The study, which analyzed human motion principles and successfully applied them to a robotic gripper, is published in the journal Science Robotics.

This research was motivated by the human grasping strategy called “multi-object grasping,” which involves transferring multiple objects at once. The team stated that the study began in 2019 after observing factory workers who, instead of moving objects individually, transferred multiple items simultaneously to enhance the efficiency of the pick-and-place process.

Professor Kyu-Jin Cho explained, “Unlike humans, conventional gripper designs have predominantly focused on transferring a single object at a time. While grippers for multi-object grasping have been developed, they typically consist of multiple small grippers arranged at the end of a robotic arm, restricting their use to highly structured environments.”

Therefore, the research team analyzed the human multi-object grasping strategy, which is applicable in unstructured environments, and developed a gripper that mimics this capability. This gripper can sequentially grasp objects, store them, transfer multiple items simultaneously, and place them individually at desired locations.

To transfer multiple objects in unstructured environments, humans grasp individual objects with their fingers and translate them to the palm for storage. This action is known as finger-to-palm translation. Finger-to-palm translation frees the fingertips to grasp new targets and can be repeated to simultaneously store multiple objects in the palm.

After transferring multiple stored objects at once, humans use palm-to-finger translation to retrieve the stored objects one by one with their fingers and place them at desired locations. These in-hand translation motions are essential for multi-object grasping in unstructured environments.

Inspired by humans’ strategy, the research team introduced a gripper with in-hand translation capability. The gripper features fingers that can translate objects and a conveyor palm for storing multiple items simultaneously. The fingers, powered by two motors, grasp objects and transfer them to the palm.

Decoupling links in the fingers kinematically separate the grasping and transferring motions, simplifying their control. The conveyor palm is designed as a pair of belts with elastic hairs embedded on their surfaces and actuated by a single motor. When the belts rotate, objects are inserted between the hairs, enabling the simultaneous storage and transfer of multiple objects.

The research team demonstrated that the developed gripper (MOGrip) can be applied to various unstructured pick-and-place processes through lab-scale experiments. In a logistics environment, they showed that eight objects on a shelf could be transferred with just two reciprocating motions. Compared to single-object grasping, this reduced process time by 34% and the robotic arm’s travel distance by 71%.

Additionally, in a domestic setting, the gripper successfully stored all objects on a desk and placed them individually at desired locations. MOGrip is expected to be applicable not only in logistics and home environments but also in other unstructured tasks, such as bin-picking processes.

Professor Cho, the senior researcher, stated, “The principles of nature provide inspiration for efficient robotic motion and actuation strategies. Finger-to-palm translation and palm-to-finger translations are a key component of the multi-object grasping strategy, and the developed gripper is the first example of applying this principle to a robot. It is expected to enable efficient pick-and-place operations in various unstructured environments, such as logistics and home settings.”

Highlighting the significance of the study, he added, “The role of roboticists is not to simply mimic natural movements but to identify core principles and adapt them to robotic applications.”