MAJOR SERIES · 10 ESSAYS

Designing the contact before designing the sensor

A robotic fingertip is often specified with a material name and a Shore hardness. That is convenient for purchasing silicone. It is not enough to explain what the fingertip will do.

The same elastomer can behave like a local force probe, a broad stabilizing contact, a mechanical low-pass filter, or an almost unreadable lump of rubber. The difference comes from geometry: the outer curvature, the rigid backing, the bonded interface, the internal stem, the voids, and the route by which deformation reaches an observable surface.

This series follows one engineering question:

Can fingertip morphology make contact more predictable and measurable before a more complicated sensor or controller is added?

The articles are based on the PLATO Hand contact design and the lit_ws fingertip framework: a parametric half-ellipse pad, solver-independent meshing, a nearly incompressible hyperelastic model, frictionless contact validation, a no-void indentation baseline, and the contact-to-observation deformation transfer map (CODTM).

The status labels matter:

• Established means the claim is supported by the current geometry, validation, or numerical results.
• Diagnostic means the result explains a failure but is not yet a production remedy.
• Research boundary means the required model or experiment has not been completed.

The series

1. A Robotic Fingertip Is Not a Rubber Cap — Why hardness alone cannot describe contact behavior.

2. The Simplest Parametric Fingertip — What a half-ellipse pad, bonded plate, stem, and two clearance parameters can control.

3. Where Does the Internal Structure Send Deformation Energy? — Why backing and load path matter as much as the elastomer.

4. Why Silicone FEM Lies So Easily — Volumetric locking, mixed elements, and why a solver return value is not validation.

5. A Converged Contact Solver Can Still Be Wrong — Active sets, force closure, mesh resolution, and physical acceptance gates.

6. A Solid Fingertip Was Easy — The external-contact-only baseline and what it actually establishes.

7. Can Contact Location Be Read from Deformation? — CODTM, sidewall signatures, distance matrices, and provisional mesh convergence.

8. Then I Added a Hole — How an internal void changes the mathematical problem, not just the compliance.

9. Mechanical Separability Is Not Sensing — Why distinguishable FEM fields do not yet imply camera observability.

10. What Should a Good Fingertip Optimize? — A research contract for combining contact mechanics, optical transfer, noise, and robustness.

The first eight articles report what the current models and diagnostics can support. Articles 9 and 10 define the next missing layer. They deliberately do not present an optical result or an optimized fingertip that does not yet exist.