Transformable Fingertip to Augment Tableware Grasp Capability

Presented in International Conference on Control, Automation and Systems (ICCAS), 2022

Abstract

This paper suggests a transformable fingertip that can make a gripper grasp tableware of various sizes and shapes. Our fingertip can enhance both the graspable range of the gripper and the grasp success rate in various tableware. A gripper design for tableware is difficult due to constraints such as limited graspable region by food on the objects and a mechanical disadvantage to making the gripper large to grip tableware of various sizes. We overcame this challenge with a novel fingertip mechanism that can rotate its fingers. The gripper can take advantage of multiple contact surfaces, the shape of which is customized for grasping objects of various widths and shapes by rotating the fingers.

Fig 1. Two fingertip modes. Mode 1 is for a narrow gripand Mode 2 is for a wide grip. The modes can be switched by fully closing grippers.

Fig 2. Grasping example of two fingertip modes.

Mode-Switching Mechanism

The mode-switching mechanism, delineated in Fig. 3, pivots the fingertip around axis (x) to utilize its different sides. Comprising two components, the first employs a button, ratchet, wire, and springs for rotating the finger, while the second uses a key to lock the finger, averting unintended mode switching. The sequence initiates by closing the gripper till buttons in Fig. 4 contact the opposite finger. Initially, button press can't rotate the ratchet axis and finger-side due to key lock, though it tensions the spring. Post this phase, the button actuates the key, unlocking and enabling rotation.

Upon button activation, a wire spins the ratchet axis, rotating the fingertip. The key returns to its original position once unlocked, inserting into the finger-side keyhole at the desired fingertip angle. Releasing the gripper resets the button and ratchet via springs ksmall and klarge, while the internal gear keeps the fingertip steady.

Fig. 4 illustrates a gripper mechanism where the button length is denoted as L and the gripper stroke as S. The design necessitates the button to be pressed when the gripper is fully closed, thereby consuming a part of the gripper stroke. This design aspect modifies the effective gripper stroke for object grasping to (S − L). If buttons on both fingers were aligned symmetrically, the real stroke would be reduced to (S − 2L). However, the buttons are designed asymmetrically to lessen the stroke needed for mode switching.

Fig 2. Schematic of mode-switching mechanism.

Fig 3. Position of the buttons according to the state transition of the gripper.

Fingertip modes

The fingertip consists of three contact surfaces labeled as (a), (b), and (c) as per Fig. 3. Mode 1 is defined by the alignment of surfaces (a) and (b) parallel to the gripper's forward direction (f). Conversely, Mode 2 is characterized when surface (c) aligns parallel to (f). The rotation of surfaces (a), (b), and (c) around axis (x) necessitates the angle between each surface and (x) to match the angle between (x) and (f) to ensure the main contact surface in each mode aligns with (f). In Mode 1, surface (a) accommodates small objects ranging 0-75mm, while (b) is suited for medium objects of 35-110mm width. Mode 2, with surface (c), caters to wide objects sized between 165-240mm, a range determined based on the largest tableware piece measured at 240mm in width.

Fig. 4. Tableware grasp experiment using Robotiq 2f-85 [16]: Conventional fingertip (v1-3, h1-3) and proposed prototype (v4-6, h4-6).We tested 3 objects of different types of shapes and sizes (fork, mug, plate). v1-v6 shows grasp tests in which direction of end effector is perpendicular to the table plane, while h1-h6 shows grasp tests when the direction of the end effector is parallel to the table plane.

Experiment and Result

A comparative analysis was conducted between the proposed prototype and a traditional parallel jaw gripper, utilizing Robotiq 2f-85 gripper and ABB IRB 120 robot for experimental purposes. The conventional fingertip was designed to match the length of the proposed version to ensure a fair comparison. The evaluation involved testing the grasp capability of both grippers on three types of tableware—forks, mugs, and plates, selected to represent a variety of tableware types and sizes commonly found in South Korea, under two distinct grasp situations: vertical and horizontal poses. The assumption was that grasp direction might be constrained if food or beverage is present in the mug and plate.

The results, illustrated in Fig. 4, were marked with a red cross for unsuccessful attempts and a green circle for successful grasps. Unlike the conventional fingertip, which could only grasp objects narrower than its stroke, making it unable to grasp the plate, the proposed prototype successfully grasped all objects in both scenarios. In Mode 1, contact surfaces (a) and (b) securely held the fork and mug, respectively, while in Mode 2, contact surface (c) accommodated the plate. Subsequent testing on stably placed tableware and comparing success rates further highlighted the superior performance of the proposed prototype.

The conventional gripper recorded a 60% and 50% success rate in vertical and horizontal grasp poses on the fork, whereas the prototype exhibited a 100% and 90% success rate, respectively. The conventional gripper's failure in certain instances was attributed to the grippable width being narrower than the target object, demonstrating the enhanced adaptability of the proposed prototype in handling a wider range of object sizes and types.