Year: 2026
Type: preprint
Abstract: A template, the simplest model that reproduces a behaviour, is a hypothesis about what the body must do. For running, the spring-loaded inverted pendulum (SLIP) captures the bounce of the centre of mass but not the force behind it: its ground reaction force (GRF) is symmetric in time and always points along the leg. Its many extensions have never faced a common test. We fitted 34 templates to stance GRF and ranked them by the Bayesian information criterion and cross-validation on six open human datasets (414 runner–speed conditions, 2–5 m s −1 , treadmill and overground) and on aerial running in two guinea-fowl datasets. Every dataset agreed. A running template needs two elements: leg damping, without which errors stayed above 0.3 body weights (BW), and a force off the leg axis, without which the fore-aft error stayed near that of SLIP. In humans, SLIP with leg damping and a hip torque that decays through stance (TD-SLIP) has both; its two-parameter form was best in the three largest datasets and within 9.3 BIC per condition of the best elsewhere (errors 0.15–0.18 BW). Guinea fowl needed the same two elements in another form, a virtual pivot above the centre of mass, and kept it on slippery ground. Parameters varied between runners far more than their gaits, and overground running softened the leg by 8% without changing the ranking: runners share the template’s structure, not its numbers. The GRF fixes the template but not how leg and hip share the work, which joint-level data must decide. Summary statement Across 34 models, six human datasets and running guinea fowl, a running template needs leg damping and a hip torque; individuals share its structure, not its numbers.
Authors: M. Mert Ankarali, Uluç Saranlı
Institution: Middle East Technical University
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7235453319
