Critique: Null Space Forces and Closed-Chain Indeterminacy

Critique and response context for Null Space Forces and Closed-Chain Indeterminacy in AffineDrift’s control-affine golf-swing framework.

Critique: Null Space Forces and Closed-Chain Indeterminacy

Summary of Concern

The framework defines “Input forces” as arising from “actively applied joint torques” (\(u\)). However, the golfer-club system (specifically the two-handed grip) forms a Closed Kinematic Chain. In closed chains, the mapping from joint torques to generalized motion is not unique; there exists a “null space” of internal torques (e.g., fighting between left and right arms, squeezing the grip) that produce zero motion and thus zero net generalized force.

Standard Inverse Dynamics (ID) resolves this indeterminacy implicitly (via pseudo-inverse/min-norm) or explicitly (via optimization). The paper defines \(\tau_{\text{input}}\) algebraically as a residual (\(\tau_{\text{total}} - \tau_{\text{drift}}\)), implying it is a unique physical quantity. This is incorrect. \(\tau_{\text{total}}\) derived from motion only captures the motion-producing component of the input. It fails to capture the internal “fight” or co-contraction, which are metabolically expensive and physically real “inputs”.

Consequently, the “Input Force” metric is a theoretical lower bound (projection), not a complete measure of the golfer’s active torque input.

Location

  • Page: articles/affine-nature-golf-swing.qmd
  • Section: sec-drift_input (Drift vs. Input Decomposition) and sec-taxonomy
  • Claim: “Input forces arising solely from actively applied joint torques… This term represents the portion of the dynamics directly caused by the golfer’s applied torques.”

Nature of the Issue

  • Unstated Assumption: That the “Active Input” is fully observable from kinematic motion.
  • Empirical Insufficiency: Neglect of internal loading in closed chains.
  • Terminological Ambiguity: Conflating “Net Motion-Producing Torque” with “Applied Joint Torque”.

Why This Is a Problem

  • Biomechanics: A golfer might be exerting massive effort (co-contraction, antagonistic arm forces) that cancels out. The AffineDrift framework would report this as “Low Input Force”.
  • Efficiency Metrics: This biases efficiency calculations. A “quiet” swing might be high-effort but high-cancellation. The framework cannot distinguish “Efficient” from “Inefficient but Isometrically Stressed”.
  • Validity: The claim of “Exact Decomposition” fails because the decomposition is fundamentally indeterminate without force sensors (e.g., grip pressure or foot reaction forces) to resolve the closed loop.

Evidence / References

  • Featherstone, R. (2008). Rigid Body Dynamics Algorithms. (Closed-loop constraints and indeterminate forces).
  • Yamaguchi, G. T. (2001). Dynamic Modeling of Musculoskeletal Motion. (Indeterminacy in muscle recruitment and closed chains).
  • Nikooyan, A. A., & Zadpoor, A. A. (2011). Mass-spring-damper modelling of the human body to study running and hopping dynamics. (Discussing indeterminacy).

Severity

  • High: It fundamentally limits the claim that the framework captures “Golfer Effort”. It only captures “Net Effective Effort”.

Suggested Remedies

  1. Explicit Qualification: Rename “Input Force” to “Net Motion-Producing Input Force” or similar.
  2. Acknowledge Internal Forces: Add a section in Limitations regarding “Null Space Torques” and “Internal Loading”. State clearly that the framework is blind to isometric exertion.
  3. Closed-Loop Constraint Formulation: If the Simulink model uses a specific method to resolve the closed loop (e.g., specific joint constraints), this must be stated. If it assumes an Open Chain (e.g., lumped arms), that assumption must be defended.
  4. Inline Edit: In Section sec-limitations (Physiological limitations), add: > “Furthermore, for closed-chain kinematic topologies (e.g., two hands on the club), Inverse Dynamics recovers only the net motion-producing torque. Internal forces (antagonistic co-contraction or forces in the null space of the motion Jacobian) are mechanically real and metabolically costly but do not appear in \(\tau_{\text{total}}\). Thus, \(\tau_{\text{input}}\) represents the net effective torque, likely underestimating the total magnitude of muscle force.”