top of page
Finger Flick Model
Simulation | Mechanical Engineering
This project mathematically simulated a human finger flick by modeling the finger as a two-link rigid-body manipulator, representing the knuckle and PIP joints. Real-world motion data was captured using Mediapipe Hands in Python, extracting 3D trajectories of key joints. The system was dynamically simulated using Lagrangian mechanics, incorporating joint limits, link inertia, and tendon-driven actuation modeled via nonlinear extension functions. Tendon forces were tuned to replicate the rapid, snapping motion characteristic of a flick.
Key Challenges
Unrealistic Oscillations: Initial simulations with high constant tendon forces produced unrealistic behavior. This was addressed by introducing decaying exponential force profiles and angular constraints to model a natural "snap."
Model Simplification: Omitting the DIP joint led to inaccuracies; resolved by enforcing a constraint (θ₃ = θ₂) to approximate its behavior while maintaining a tractable two-link model.
Biomechanical Fidelity: Differences in real vs. simulated motion due to elastic tendon behavior were reduced by iteratively refining force profiles and improving joint mass modeling.
More Resources
Project Paper
bottom of page