TAG Results

With TAG designed and fabricated, its performance was then tested. First, the final TAG finger designs were compared again to a more traditional design to ensure that the FEA findings still held even when the Fin Ray geometry was further modified. Then the load capacity of each gripper was tested. Finally, several grasping tests were performed, including picking up objects, navigating environments, and opening doors.

Load Capacity

The TAG gripper was capable of holding approximately 40% more than the gripper with traditional fingers. The TAG gripper was capable of holding an average of 560 ± 10 g, while the traditional gripper was able to hold an average of 390 ± 10 g (Figure 3.5). The difference was analyzed using a t-test and is statistically significant (p << 0.001).

During the lateral trials, the TAG was capable of holding 430 ± 27 g but a load capacity for the gripper with the traditional fingers could not be determined because the joints began to separate from the motor holder atloads of approximately 300 g. Separation occurred to a lesser extent in the TAG but did not prevent data collection, and separation did not occur in the vertical trials, but since the gripper will need to be able to pick up objects in either orientation, improvements will need to be made to the gripper/motor holder interface to overcome this issue.

As expected from the modeling results, the TAG possesses greater ability to conform to objects results in it being able to carry a greater load than the traditional gripper. Both grippers were printed out of the same material and thus were subject to the same material strength limitations. Although printed from the same material, the TAG gripper outperformed the traditional gripper by 15%. This supports the hypothesis that the modifications made to the Fin Ray Effect finger allowed the TAG gripper to hold more weight because it had a more complete and stable grip than the traditional gripper.

Due to the failure of the grippers to complete the lateral load experiments, the interface between the joints and the motor holder should be redesigned to prevent separation and decrease deformation under lateral loads. Alternatively, the gripper could be printed or molded from materials with a higher resistance to tear.

Forces Exerted

Discrete force readings of the TAG were recorded and compared to the FEA models for three different applied forces. On average, the normal forces exerted that were measured experimentally were approximately 20% larger than those determined by FEA (0.75 ± 0.13 N compared to 0.62 N). This difference is likely due to the addition of hard crossbeams to the gripper that were not represented in the material model, which was composed of silicone. The inclusion of hard components would enable the gripper to produce larger normal forces than the TAG would have otherwise been able to produce.

Continuous force readings for the TAG were recorded, and two representative samples (left) are depicted. The first image shows that crossbeam 3 makes contact with the pipe first followed by crossbeams 1 and 5, which make contact at almost the same time with one sometimes making contact before the other. The normal force at crossbeam 1 increases as the applied force increases and decreases as the applied force decreases, but the normal forces at crossbeams 3 and 5 increase, peak, and decrease as the force is applied and increase, peak, and decrease as the applied force is decreased. The bottom left image shows a similar result with the crossbeams 2 and 4 acting like 3 and 5 and crossbeam 6 acting like crossbeam 1.

The interesting finding here is that the TAG crossbeams that exert the largest force are not solely at the base but also at the tip, with the smallest forces occurring in the middle. Prior to performing these experiments, it was expected that the largest forces would occur closer to the base, and this is, indeed, what the mathematical model predicts. However, this U- shaped force distribution might be what makes robotic fingers an excellent application for the Fin Ray Effect. Because the TAG is able to exert such a large for at the tip, just like human fingers, objects are less likely to slip from its grasp.

RoboSoft Grand Challenge

From Table to the left, it can be seen that TIHRA performed well in some portions of the manipulation challenge but not in others. The successes of the gripper included the fingernail and its ability to conform around objects. The fingernail enabled the gripper to slide along a surface in to pick up an object. Much like human fingernails, TIHRA’s fingernail was able to get under thin objects, like the small, thin wrench, and pull them away from the surface before closing fully to grasp the object. The gripper was also able to conform to objects, like the ice cream cone, without damaging them. Additionally, the success in picking up the ice cream cone highlighted the gripper’s ability to handle delicate and fragile objects without breaking them, which is an area of di culty for hard robotic grippers.

A major source of poor performance in the challenge was not the gripper itself but rather the arm design. While the arm enabled the gripper to reach many of the objects, it was often impossible to orient the gripper into a configuration ideal to complete the task. Because of this, either the gripper could not close fully around the object (e.g. the water bottle) or the gripper could not reach an object from a favorable angle (e.g. the Tupperware). Additionally, the torque supplied by the motors was not enough to overcome lifting the manipulator with large load, especially when lifting the arm from certain configurations. These issues resulted in unstable grips and poor motion.

The TAG’s poor performance can also be partially be attributed to its lack of ability to exert the force required to complete a task. When the TAG attempted to lift the water bottle, the lateral deformation of the gripper as well as slipping from the grip caused the gripper to drop the object. This was exacerbated by an incapability to position the arm so that the gripper was in an ideal orientation with respect to the object as well as an incapability of the arm to lift the additional load. When tasked with opening the door, TAG deformed by twisting and could not provide enough force to turn the handle. Improving the gripper to better withstand lateral and torsional forces as well as attaching it to an arm with more degrees of freedom would help solve these problems.