19.3 - Initial Prototype

We use physical prototyping as a part of our design process to validate and refine the compliant finger mechanism. By building and testing, we aim to identify problems that are not always found in simulation. This may include assembly, material behavior, and the actuation performance. This process is iterative in which each new prototype will inform improvements for the next design. Our prototyping documentation is outlined below.

MotionGen Model

To begin our prototyping efforts, we developed a digital model using MotionGen to simulate the finger mechanism. This allowed us to approximate the desired motion profile and scale for the system. The preliminary model is shown below.

Open

Figure 1. Preliminary MotionGen Model to Guide Link Lengths and Scale

From this model, we retrieved approximate link lengths to guide our first physical prototype. The links are defined as follows:

The joint between Link 3 and Link 4, denoted by the plus sign in MotionGen, is where the torsional spring is located.

Note that, while the MotionGen model visually depicts the links with extra joints to create their unique geometries, all links are treated as binary links for the purposes of our project. The sketch below shows the links using their defined link numbers and without their geometric forms.

Open

Figure 2. Simplified Kinematic Diagram to Show Binary Links, Not to Scale

Initial CAD from MotionGen Model

From the MotionGen model, we created our initial CAD model using the approximated link lengths. The screenshots below depict the first version of our compliant finger mechanism.

Open

Figure 3. Isometric View of CAD Finger Assembly

Open

Figure 4. Top View of CAD Finger Assembly

However, during the assembly simulation, links 3 and 5 unexpectedly collided, due to the geometry of link 5. This restricted the finger from closing properly, so the geometry of links 4 and 5 were altered. The length of link 4 was also changed to 15 mm. These changes are reflected in the images below.

Updated Model to Avoid Collision and add Torsional Spring

Open

Figure 5. Isometric View of CAD Finger Assembly

Open

Figure 6. Top View of CAD Finger Assembly

In these CAD models, slots were also added to links 3 and 4 to pre-emptively account for the torsional spring’s integration. While at this time, the springs have not been ordered or received, we wanted to implement this design update early such that we can begin iteration as soon as the components arrive. The slot for the torsional spring can be seen in the CAD screenshot below.

Open

Figure 7. Torsional Spring Slot in Link 3

Updated Model Through Preliminary 3D-Prints

Through 3D-printing the updated CAD models, we identified a new issue. Although the finger appeared to bend as intended in simulation, the physical assembly (Figure 8) revealed that the distal segment bent excessively even when the proximal segment was only roughly perpendicular to the base.

To address this, we increased the length of link 3 from approximately 7 mm to 15 mm to limit distal over-flexion in the first stage of motion (Before Object-Contact/Unflexed Torsional Spring). With this modification, the updated assembly (Figure 9) shows that the distal segment remains appropriately aligned while the proximal segment maintains a similar position.

Open Figure 8. Short Link 3 Creates Unwanted, Early Distal Bend	Open Figure 9. Longer Link 3 Prevents Early Distal Bending

Open Figure 8. Short Link 3 Creates Unwanted, Early Distal Bend	Open Figure 9. Longer Link 3 Prevents Early Distal Bending

Furthermore, as seen in the above images, we chose to increase the base of our finger mechanism. This allows the base to provide another area of contact with the object being grasped, assisting the finger mechanism naturally as a palm would.

Updated Finger Geometry and Finalized Link Lengths

To evaluate the grasping capability of our compliant finger mechanism across objects of varying sizes, we selected a “watch cushion” as our test object. This cushion is a small rectangular block with rounded edges, measuring approximately 3” × 1.5” × 1.25”. Its geometry allows it to be oriented along its longer or shorter dimension, enabling us to test the mechanism under different grasping conditions. The test object is shown in Figure 10.

Open

Figure 10. Watch Cushion as our Test Object

Initial testing showed that the mechanism successfully grasped the object when oriented along its longer dimension. In this case, the system remained in its stage one four-bar configuration, and both proximal and distal segments made contact with the object effectively. However, we encountered an issue when testing the shorter orientation (approximately 1.5 inches). Here, the mechanism transitioned into the stage two four-bar behavior, where the torsional spring is expected to flex and allow the distal segment to wrap around the object.

Despite this intended behavior, links 2 and 3 came into contact with the object, preventing the distal segment from achieving a secure grasp. This limited the effectiveness of the mechanism when grasping smaller objects.

Open

Figure 11. Object is not tightly grasped, due to link geometry

To address this issue, we experimented with MotionGen to simulate different changes. Based on these iterations, we reduced the length of link 4 from 15 mm to 10 mm. Additionally, we increased its thickness, allowing the distal segment to engage the object earlier in the grasping motion. A comparison between the original (black) and updated (purple) link 4 designs is shown below.

Open

Figure 12. Link 4 Update (Old in black, new in purple)

This resulted in an improved performance, enabling a more reliable and tighter grasp as seen in Figure 13.

Open

Figure 13. Better, tighter grasp with updated link geometry

Prototype Results

Through iterative design and testing, we updated the mechanism to improve its functionality. The final link dimensions were selected based on these improvements and are summarized below.

Although the final prototype does not yet include torsional springs, the mechanism’s intended behavior is still demonstrated. When grasping larger objects, the finger operates as a collective system in its stage one configuration, with both proximal and distal segments making contact simultaneously (Figure 14). For smaller objects, the proximal segment contacts the object first, effectively grounding that link. This triggers the transition to stage two behavior, where the torsional spring would allow continued motion of the distal segment, enabling it to wrap around and securely grasp the object (Figure 15).

Open Figure 14. Finger grasps a larger object (Spring is unflexed)	Open Figure 15. Finger grasps a smaller object (Spring is flexed)

Open Figure 14. Finger grasps a larger object (Spring is unflexed)	Open Figure 15. Finger grasps a smaller object (Spring is flexed)

Prototype Bill of Materials

While our prototype only entails the development of a single compliant finger mechanism to demonstrate its unique motion profile, we use a variety of materials and off-the-shelf components to create a working assembly. The following draft BOM documents the parts, relevant dimensions, quantities, prices, and sources that make our prototype come to life.

Note that for this BOM, we list electronic components along with the torsional springs as we plan to implement these for testing beyond manual actuation. However, due to delays in shipping and the requirements of our prototype demonstration, both the spring and electronics are not yet physically implemented into our assembly. These items are italicized in the BOM for clarity.

From this draft BOM, we can see that our prototype primarily uses available parts from our class’s resources and from TIW, keeping our total costs relatively low.

Plans to Transition from Prototype to Final

With the current prototype complete, we have identified several key improvements for the final iteration of our project. First, we will integrate the torsional spring into the mechanism and conduct testing to validate its functionality.

Secondly, we plan to refine the link designs to enhance contact surfaces, making them more anatomically finger-like while also improving durability. These updates aim to achieve more effective and reliable grasping.

Lastly, we will expand the system from a single finger to a multi-finger configuration, targeting a total of two to three fingers to form a simplified robotic hand. To maintain design efficiency, all fingers will be actuated using a single motor. We anticipate implementing either a gear-based transmission system, belts, or a linkage mechanism to distribute the rotary input across multiple fingers.

With these changes, the final system will demonstrate a functional, simplified hand composed of compliant finger mechanisms capable of adaptive grasping across different object sizes.