20.6 - Conclusions and Future Work

Conclusions

This project successfully achieved its primary objective of designing, modeling, and validating a tendon-actuated compliant robotic finger using the pseudo-rigid-body (PRB) approach. The PRB model proved to be an effective tool for approximating the behavior of the compliant TPU structure, allowing us to apply classical kinematic analysis techniques to a system that would otherwise be difficult to model.

Through the development of the analytical model, we established clear relationships between actuator input (tendon displacement), joint stiffness, and resulting finger motion. The model was able to predict fingertip trajectories, joint angles, and general motion trends with reasonable accuracy when compared to experimental observations from the physical prototype.

A key outcome of this work was the successful fabrication and integration of a motor-actuated compliant finger. Iterative prototyping allowed us to identify and address important design tradeoffs, particularly between compliance and structural stiffness. The transition from a manually actuated prototype to a motor-driven system significantly improved repeatability and demonstrated the feasibility of controlled actuation.

By the end of the project, we expanded the system into a two-finger planar gripper. This final implementation demonstrated the ability to grasp and conform to objects of varying shapes, validating the intended advantage of compliant mechanisms: increased contact area and safer interaction with objects. The gripper exhibited the desired curling behavior and showed evidence of sequential joint engagement, although not perfectly isolated, confirming that tendon routing and joint stiffness distribution strongly influence motion coordination.

Overall, the project demonstrated that PRB modeling is a practical and efficient method for designing compliant robotic fingers with predictable behavior, bridging the gap between purely empirical design and computationally intensive simulation.

Lessons Learned

One of the most important lessons learned was the sensitivity of compliant mechanisms to small geometric and material changes. Minor adjustments to flexure thickness, tendon routing, or material stiffness had significant effects on both motion and force output.

We also learned that friction plays a much larger role than initially expected. Tendon routing through small slots introduced losses that impacted both actuation efficiency and the finger’s ability to return to its original position. This highlighted the importance of considering real-world non-idealities alongside theoretical models.

Another key takeaway was the limitation of simplified models. While the PRB approach provided valuable insight, it does not fully capture distributed deformation or contact interactions. As a result, discrepancies between predicted and actual motion were inevitable, especially at higher deflections.

Finally, iterative prototyping proved essential. Many of the most critical insights – such as tendon material selection and stiffness tradeoffs – could not have been identified through analysis alone.

Future Work

While the current two-finger gripper demonstrates successful operation, several improvements could elevate the design to the next level.

One major area for improvement is the refinement of tendon routing. Reducing friction through smoother channels, larger clearances, or low-friction liners would improve efficiency and repeatability. Additionally, incorporating return mechanisms, such as embedded elastic elements or antagonistic tendons, would improve the finger’s ability to reopen reliably.

The PRB model could also be extended to include more advanced effects, such as nonlinear stiffness, frictional losses, and contact forces. Incorporating these factors would improve predictive accuracy and better align the model with experimental behavior.

From a hardware perspective, integrating sensors – such as encoders for joint angle estimation or force sensors at the fingertip – would allow for closed-loop control and more precise manipulation. This would be especially valuable for handling delicate or irregular objects.

Expanding beyond a planar gripper is another natural progression. Adding additional fingers or arranging them in a three-dimensional configuration would enable more complex grasping strategies and improve overall dexterity.

Additionally, expanding the hardware control system to allow for individually actuated fingers would give the device more control during complex operations. While our team attempted to achieve this with a stacked NEMA 17 motor the torque was too low but could be improved with gearing or material re-selection.

Finally, future iterations could explore alternative materials or multi-material 3D printing to better tune stiffness distribution along the finger, enabling more controlled sequential curling behavior.

Tips for Future Groups

Start prototyping early, even with simple designs, as physical testing reveals issues that are not obvious in analysis.

Be mindful of friction in tendon-driven systems, as it can dominate system behavior if not properly managed.

Use parametric CAD models to quickly iterate on geometry and directly link design changes to analytical models.

Do not rely solely on theoretical models – combine modeling with experimentation for best results.

Acknowledgements

We would like to thank TIW for providing access to fabrication resources and guidance throughout the project. We also appreciate the support from Dr. Symmank who provided feedback during the design and testing phases. Her input was valuable in refining both the physical design and analytical approach.

Sources

[1] L. L. Howell, Compliant Mechanisms. New York, NY, USA: Wiley, 2001.

[2] J. Sun, C. Chen, L. Wang, et al., “Design and simulation experiment of rigid-flexible soft humanoid finger,” Machines, vol. 10, no. 6, p. 448, 2022.

[3] Y. Zhang, W. Zhang, J. Yang, and W. Pu, “Bioinspired soft robotic fingers with sequential motion based on tendon-driven mechanisms,” Soft Robotics, 2022.