Central pivot point control of a body-attached spring-mass runner: periodic gaits, stabilization, template embedding and approximate analytic return maps

İzel Sever, O. Kaan Karagöz, M. Mert AnkaralıarXiv 2026 · arXiv

TL;DR

This paper introduces the Central Pivot Point (CPP) control law, which restores the elegant central-force structure of simple spring-mass models to complex runners with trunks. By decoupling body posture from center-of-mass motion, the authors derive the first approximate analytic return maps for trunk models, enabling fast, precise, and stable "deadbeat" control for legged robots.

What Problem Does This Paper Address?

The Spring-Loaded Inverted Pendulum (SLIP) is the "gold standard" template for understanding running, but it lacks a trunk. Real robots and animals have heavy bodies that must be balanced.

In traditional models, the hip torques required to maintain trunk posture create non-radial forces. This "couples" the trunk's orientation (pitch) with the body's center-of-mass (COM) movement, destroying the "central-force" property that makes SLIP mathematically tractable. Consequently, researchers have had to rely almost entirely on slow, black-box numerical simulations to study runners with trunks. This paper aims to restore the analytic tools of SLIP to these more realistic models.

Methods

The core innovation is the Central Pivot Point (CPP) control law. The authors propose a specific hip torque that ensures the ground reaction force (GRF) always points directly through the COM (see Figure 1).

FIG. 1. (a) Trunk-SLIP (TSLIP) model and its parameters; the labels dhip and d correspond to dh and c in the text. (b) Central pivot point (CPP): the hip torque τ adds a force component Fτ = τ/r normal to the leg, so that the GRF passes through the COM. (c) Virtual leg from the toe to the COM, with length rv and angle θv

Key technical components include:

  • Decoupling: Under CPP, the trunk pitch becomes a "pure double integrator," meaning it can be controlled independently of the bouncing motion.
  • Template Embedding: Using Partial Feedback Linearization (PFL), the authors embed a "linear template" where the radial leg motion behaves like a simple harmonic oscillator.
  • Analytic Return Map: They derive a closed-form mathematical map to predict the state of the next "hop" (apex) using a novel linear-moment gravity correction (GC-L).
  • Hierarchical Control: A continuous PD controller handles trunk pitch during stance, while a stride-level "deadbeat" controller adjusts touchdown angle and leg thrust to hit target speeds and heights.

Datasets

The study does not use external datasets. Instead, the authors validated their approach using extensive numerical simulations of a human-sized model:

  • A grid of 10,977 periodic gaits was computed to characterize the system's natural behavior.
  • Over 70,000 individual strides were simulated to test the accuracy of the analytic return maps across various speeds, heights, and postures.

Key Insights and Ideas

  • Posture-Invariant Embedding: Because of the decoupling, a robot can lean forward (to carry a load or accelerate) or backward without changing the underlying "bouncing" dynamics of its gait.
  • Symmetry and Gait Families: At an upright posture, the system exhibits time-reversal symmetry. This means periodic gaits aren't just isolated points; they exist in continuous "families," allowing for smooth transitions between different running speeds.
  • The "Natural" Gait: The authors identified a set of "natural gaits" where the force step at touchdown—a common physical unrealistic artifact in active control—completely vanishes.
  • Energy Injection via Leaning: Interestingly, leaning forward naturally injects energy into the stride, while leaning backward removes it, providing an intuitive way to manage speed (see Figure 4).
FIG. 3. One stride of the passive CPP runner from the upright fixed point (αtd = 109.8◦) with initial pitch θ0 ∈{0◦, 5◦, −5◦} and zero pitch rate. The pitch stays constant (not shown). Forward lean injects energy and backward lean removes it. The shaded interval marks stance

Key Results

  • High Accuracy: The new analytic map (GC-L) predicts liftoff velocity with a mean error of 1.6% and apex position with just 0.9% error.
  • Extreme Speed: The closed-form map computes in microseconds, making it hundreds of times faster than numerical integration and suitable for real-time model-predictive control.
  • Robust Recovery: The hierarchical controller successfully recovers from massive disturbances—such as a 20° pitch error combined with a 100°/s pitch rate—even when realistic torque and friction limits are applied (see Figure 6).
The controller recovers from large pitch and velocity disturbances within just a few strides.

Limitations and Future Work

  • Massless Legs: The model assumes legs have no mass, ignoring the complex dynamics of swinging a heavy leg forward during flight.
  • Instantaneous Placement: Leg placement is assumed to be instantaneous, which isn't possible for physical hardware.
  • Actuation Requirements: The embedding requires an actuator capable of producing force steps and managing energy flow, which may be challenging for some motor designs.
  • Simulation Only: While the results are mathematically rigorous, they have only been tested in simulation. The next logical step is implementation on a physical bipedal or quadrupedal robot.

Summary updated

Abstract

The spring-loaded inverted pendulum (SLIP) describes the center-of-mass (COM) motion of runners but has no trunk. In trunk models, the hip torque couples posture and COM motion and destroys the central-force structure on which SLIP analysis relies. As a result, trunk models have been studied almost only numerically. We show that this structure can be restored. The central pivot point (CPP), a hip-torque law that directs the ground reaction force through the COM, decouples pitch exactly and reduces stance to a planar central-force problem. At upright posture, time-reversal symmetry produces one-parameter families of periodic gaits. A stance-phase pitch law and a stride-level regulator stabilize all computed gaits, also with bounded hip torque and friction. Under CPP, any radial force law in virtual-leg coordinates can be embedded independently of posture. For a linear radial template, the stance radius and the constant-momentum sweep have exact closed forms. With a new gravity correction, the resulting analytic return map, to our knowledge the first for a trunk model, predicts liftoff velocity and apex position with mean errors of 1.6% and 0.9%. The natural gaits of the embedded runner are symmetric and need no force step at touchdown. A near-reversibility argument explains their unit multiplier, and a scalar equation predicts them within 0.8%. A deadbeat stride controller based on the analytic map tracks changes of speed, height and posture. CPP thus makes trunk models accessible to the analytic tools developed for SLIP.
BibTeX
@misc{sever2026central,
  author = {İzel Sever and O. Kaan Karagöz and M. Mert Ankaralı},
  title = {{Central pivot point control of a body-attached spring-mass runner: periodic gaits, stabilization, template embedding and approximate analytic return maps}},
  year = {2026},
  eprint = {2611.02564},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.02564},
}
Figure