Generalized Omega Turn Gait Enables Agile Limbless Robot Turning in Complex Environments

Tianyu Wang, Baxi Chong, Yuelin Deng, Ruijie Fu, Howie Choset, Daniel I. Goldman

TL;DR

This paper significantly advances the "omega turn"—a highly agile, bio-inspired turning gait for limbless (snake) robots modeled after the nematode worm C. elegans. By leveraging geometric mechanics and compliant control, the authors generalize this two-wave gait template to allow precise turn-angle modulation, adapt to robots with fewer joints, and robustly navigate complex, unstructured terrains like granular media and rock piles where traditional turning strategies fail.

What Problem Does This Paper Address?

While limbless robots excel at navigating confined spaces that wheeled or legged robots cannot reach, reorientation (turning in place) remains a surprisingly understudied challenge. Most existing turning strategies for snake robots, such as the "offset turn" (adding a constant bend to a standard slithering wave), struggle in dense clutter or loose environments like sand. If a snake robot gets wedged between rocks or pegs, rigid open-loop turning gaits often cause it to jam or fail to rotate. The authors tackle the need for a highly maneuverable, tunable, and robust turning strategy that works across various robot body lengths and unpredictable environments.

Methods

The researchers build upon their previously developed "omega turn" template, which superimposes two coplanar traveling waves: a forward wave and an omega (turning) wave. This creates an $\Omega$-like body shape where the robot's head sweeps near its tail. To generalize and improve this gait, they introduce three main technical advancements:

  1. Geometric Mechanics for Gait Optimization: They map the robot's shape variables into a 4D shape space. By calculating "infeasible regions" (where the robot would collide with itself or exceed joint limits), they optimize the gait path to maximize rotational displacement per cycle.
  2. Parameter Modulation: They systematically vary the joint angle limits (amplitude) and the spatial frequency (number of waves on the body) of the template. This allows them to precisely dial in the desired turning angle and adapt the gait for robots with fewer internal degrees of freedom (DoFs).
  3. Compliant Control: To handle unmodeled obstacles, they apply an admittance control framework to the wave amplitudes. By treating the wave parameters as a virtual spring-mass-damper system driven by proprioceptive joint torque feedback, the robot's body can physically "yield" and deform when it pushes against an obstacle, preventing jamming.

Datasets

Not applicable. This is a robotics hardware and simulation paper; it does not rely on standard machine learning datasets. Evaluation was conducted via numerical simulations (using quasi-static assumptions and Coulomb friction models) and robophysical experiments using a 16-joint limbless robot tracked via an OptiTrack motion capture system.

Key Insights and Ideas

  • Decoupling waves unlocks maneuverability: By separating the forward wave from the turning wave, the robot can maintain the locomotive benefits of standard slithering while executing massive, in-place rotational sweeps.
  • Shorter robots just need a frequency shift: A key requirement of the omega turn is that the robot's head must sweep close to its tail. For robots with fewer joints (shorter bodies), the authors found they could simply lower the spatial frequency of the omega wave to achieve the necessary local body curvature, proving the gait is hardware-agnostic.
  • Compliance is mandatory for clutter: In obstacle-rich environments, rigid gaits fail because the robot gets wedged. Allowing the mathematical wave amplitude to dynamically shrink in response to physical torque allows the robot to "slither through" the resistance rather than fighting it.
  • Standard gaits fail on loose terrain: Traditional offset and geometric turns rely heavily on stable ground contacts. On granular media (like sand or plastic spheres), they just thrash in place. The omega turn's sweeping motion, however, successfully generates rotation even when the ground gives way.

Key Results

  • Massive Tunable Range: By modulating the joint angle amplitude between $60^\circ$ and $90^\circ$, the omega turn achieved a continuous turning range of $88^\circ$ (from $20^\circ$ to $108^\circ$ per cycle). This is roughly twice the range of traditional offset or geometric turns.
  • Granular Media Dominance: On a bed of 6mm plastic spheres, the omega turn achieved $78.5^\circ$ of angular displacement per cycle. In stark contrast, the offset turn ($2.9^\circ$) and geometric turn ($4.8^\circ$) were rendered entirely ineffective.
  • Obstacle Navigation: In a pegboard environment, applying compliant control increased the robot's angular displacement by ~30% compared to the rigid, open-loop omega turn, while significantly reducing variance (fewer jams).
  • Real-World Robustness: In outdoor field tests on a pile of rocks, the omega turn maintained a highly robust $100.1^\circ \pm 6.4^\circ$ turn per cycle, nearly matching its ideal flat-ground performance ($108^\circ$).

Limitations and Future Work

Limitations: The numerical simulations rely on quasi-static assumptions, which may not perfectly capture highly dynamic movements or complex 3D slip in granular media. Additionally, while the gait generalizes to robots with fewer joints (down to 6 yaw joints tested), the maximum turning performance does drop modestly as the body gets shorter.

Future Work: The authors plan to test the omega turn in even more challenging, fully 3D environments. Crucially, they aim to develop smooth, algorithmic transitions between forward locomotion gaits and the omega turn, which is a necessary next step for integrating this behavior into higher-level autonomous motion planners for search-and-rescue or inspection tasks.

Summary updated

Abstract

Reorientation (turning in plane) plays a critical role for all robots in any field application, especially those that in confined spaces. While important, reorientation remains a relatively unstudied problem for robots, including limbless mechanisms, often called snake robots. Instead of looking at snakes, we take inspiration from observations of the turning behavior of tiny nematode worms C. elegans. Our previous work presented an in-place and in-plane turning gait for limbless robots, called an omega turn, and prescribed it using a novel two-wave template [1]. In this work, we advance omega turn-inspired controllers in three aspects: 1) we use geometric methods to vary joint angle amplitudes and forward wave spatial frequency in our turning equation to establish a wide and precise amplitude modulation and frequency modulation on omega turn; 2) we use this new relationship to enable robots with fewer internal degrees of freedom (i.e., fewer joints in the body) to achieve desirable performance, and 3) we apply compliant control methods to this relationship to handle unmodelled effects in the environment. We experimentally validate our approach on a limbless robot that the omega turn can produce effective and robust turning motion in various types of environments, such as granular media and rock pile.
BibTeX
@inproceedings{wang2022generalized,
  author = {Tianyu Wang and Baxi Chong and Yuelin Deng and Ruijie Fu and Howie Choset and Daniel I. Goldman},
  title = {{Generalized Omega Turn Gait Enables Agile Limbless Robot Turning in Complex Environments}},
  booktitle = {IEEE International Conference on Robotics and Automation},
  year = {2022},
  url = {https://openreview.net/forum?id=SZYT1rTNROv},
}

Loading related content...

Figure