16 Ways to Gallop: Energetics and Body Dynamics of High-Speed Quadrupedal Gaits

Yasser G. Alqaham, Jing Cheng, Zhenyu Gan · 2025

Read 16 Ways to Gallop: Energetics and Body Dynamics of High-Speed Quadrupedal Gaits by Yasser G. Alqaham, Jing Cheng, Zhenyu Gan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-13716.txt.

Introduction

Galloping animals save energy in ways engineers barely understand, yet robots now gallop too. Yasser Alqaham, Jing Cheng, and Zhenyu Gan sort galloping styles in 2025 using an old framework based on when feet strike and when the body floats. They model a Unitree A1 robot as a hybrid dynamic system. Trajectory optimization then finds the cheapest stride at each speed, measured by cost of transport. Sixteen styles come under the lens.

Two famous footfall orders, rotary and transverse, cost the same energy despite different body sway. What matters is floating: no flight phase wins at low speed, while two flight phases win when speed climbs. A separate controller in simulation confirms the pattern. The authors aim the results at robots that switch gaits to save power.

Catalog record

CREATORYasser G. Alqaham, Jing Cheng, Zhenyu Gan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-13716

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

16 Ways to Gallop: Energetics and Body Dynamics of High-Speed Quadrupedal Gaits Yasser G. Alqaham, Jing Cheng, Zhenyu Gan Abstract Galloping is a common high-speed gait in both animals and quadrupedal robots, yet its energetic characteristics remain insufficiently explored. This study systematically analyzes a large number of possible galloping gaits by categorizing them based on the number of flight phases per stride and the phase relationships between the front and rear legs, following Hildebrand’s framework for asymmetrical gaits. Using the A1 quadrupedal robot from Unitree, we model galloping dynamics as a hybrid dynamical system and employ trajectory optimization (TO) to minimize the cost of transport (CoT) across a range of speeds. Our results reveal that rotary and transverse gallop footfall sequences exhibit no fundamental energetic difference , despite variations in body yaw and roll motion. However, the number of flight phases significantly impacts energy efficiency : galloping with no flight phases is optimal at lower speeds, whereas galloping with two flight phases minimizes energy consumption at higher speeds. We validate these findings using a Quadratic Programming (QP)-based controller, developed in our previous work, in Gazebo simulations. These insights advance the understanding of quadrupedal locomotion energetics and may inform future legged robot designs for adaptive, energy-efficient gait transitions. I Introduction Quadrupedal locomotion has long fascinated researchers in both biology and robotics due to its complexity and adaptability. Among various gaits, galloping stands out as a high-speed strategy characterized by asymmetrical footfall patterns and aerial phases, enabling rapid and agile movement. Understanding the mechanics and energetics of galloping is crucial for developing quadrupedal robots capable of efficient and dynamic locomotion. Galloping is prevalent among mammals such as cheetahs, horses, and dogs. This gait follows a distinct footfall sequence, often incorporating one or more aerial phases per stride. As illustrated by Fig. 1 , Hildebrand’s seminal work categorized asymmetrical gaits based on limb phase relationships, distinguishing between transverse gallops , where the hind limb contacts are followed by the ipsilateral forelimb, and rotary gallops , which exhibit a circular limb sequence [ 1 , 2 ] . These patterns affect speed and stability; for instance, cheetahs employ rotary gallops for sprinting, while horses often use transverse gallops for endurance running [ 3 ] . The number of flight phases significantly influences energy expenditure [ 4 , 5 ] . Galloping with no flight phases is more energy-efficient at lower speeds, while incorporating one or two flight phases enhances efficiency at higher speeds. This adaptability allows animals to optimize gait selection based on environmental and metabolic constraints. Fig. 1: (a) Footfall sequence variations between transverse and rotary galloping gaits, with foot placement color-coded: right-fore foot (blue), left-fore foot (green), right-hind foot (yellow), and left-hind foot (red). (b) Classification of galloping gaits based on suspension phases, including no suspension, gathered suspension, extended suspension, and dual suspension [ 1 ] . Animal images sourced from [ 6 ] . Translating biological galloping into robotics presents significant challenges. Early efforts focused on walking and trotting, but achieving high-speed, agile locomotion required exploring galloping gaits. The MIT Cheetah robot demonstrated stable galloping using model-based control, reaching speeds of 6 m/s [ 7 , 8 ] . Another example, the Scout II quadruped, implemented a rotary gallop using passive dynamics and minimal sensory feedback [ 9 ] , highlighting the role of mechanical design in gait stability. Recent advances have further improved robotic speed and agility. The Black Panther II, developed by Zhejiang University, achieves 10.4 m/s while sprinting 100 meters in under 10 seconds [ 10 ] . Similarly, the HOUND robot, developed by KAIST, set a Guinness World Record with a 100-meter sprint time of 19.87 seconds [ 11 ] . These developments demonstrate the growing capabilities of quadrupedal robots. However, both robots use trotting, an intermediate-speed gait that prioritizes stability over maximal velocity. In contrast, animals rely on galloping to reach their highest speeds, emphasizing the need for further research into energy-efficient galloping gaits for robotic quadrupeds . Despite advances in robotic locomotion, replicating efficient galloping remains challenging. Maintaining dynamic stability at high speeds requires precise coordination of limb movements and body dynamics, necessitating robust control strategies. Additionally, robotic galloping often incurs a higher cost of transport (CoT) than biological counterparts, highlighting the need for optimization in both control algorithms and mechanical design. The number of flight phases plays a critical role in energy efficiency. While increased aerial phases enable higher speeds, they also require greater energy input. Optimizing this balance is essential for achieving both speed and efficiency. Furthermore, transitioning between trotting and galloping in unstructured environments introduces additional challenges, as terrain variability affects gait stability. This work systematically analyzes galloping gaits in quadrupedal robots, modeling galloping dynamics as a hybrid system and employing trajectory optimization to minimize CoT across different speeds. We categorize galloping gaits based on the number of flight phases and investigate their impact on energy efficiency. Our key contributions include: (i) Comprehensive gait analysis : A detailed examination of transverse and rotary galloping sequences and their energetic profiles. (ii) Optimization framework : Implementation of trajectory optimization to identify gait parameters that minimize energy consumption, informing efficient locomotion strategies for robots. This study bridges the gap between biological insights and robotic applications, advancing the development of quadrupedal robots capable of high-speed, energy-efficient locomotion. II Methods This section outlines the methodology adopted in this study. First, we describe the different types of galloping gaits considered. Next, we present the modeling of the quadrupedal robot A1, followed by an overview of its galloping hybrid dynamics. We then detail the formulation of the nonlinear trajectory optimization problem. II-A Footfall Sequence of Galloping Gaits Fig. 2: Gait diagrams of representative galloping gaits over one normalized stride, from 0% (beginning of the stride, defined by right front leg touchdown) to 100% (end of the stride). The footfall pattern of a gait is uniquely defined by its footfall sequence, represented as \mathbf{\boldsymbol{t}}=[t^{\text{TD}}_{\scriptscriptstyle\text{LH}},t^{\text{LO}}_{\scriptscriptstyle\text{LH}},t^{\text{TD}}_{\scriptscriptstyle\text{LF}},t^{\text{LO}}_{\scriptscriptstyle\text{LF}},t^{\text{TD}}_{\scriptscriptstyle\text{RF}},t^{\text{LO}}_{\scriptscriptstyle\text{RF}},t^{\text{TD}}_{\scriptscriptstyle\text{RH}},t^{\text{LO}}_{\scriptscriptstyle\text{RH}}]^{T} . Each timing value t^{j}_{i} (where i\in\{\text{LH},\text{LF},\text{RF},\text{RH}\} and j\in\{\text{TD},\text{LO}\} ) is normalized by the stride duration t_{\text{stride}} and ranges from 0 to 100%. The initials L, R, F, and H represent left, right, fore, and hind feet, respectively, while TD and LO denote touchdown and lift-off events. Using Hildebrand’s convention [ 1 ] , footfall sequences are illustrated with black bars for ground contact and white spaces for suspension, with total stride duration spanning the full sequence (e.g., Fig. 2 ). To quantify different galloping gaits, four phase-based metrics are introduced: Fig. 3: Illustration of different variations of galloping gaits. Panels (a)–(d) depict gaits with varying numbers of flight phases (shaded in gray). Each gait is shown under two primary gallop styles: transverse (top) and rotary (down). The subplot on the right represents the stride cycle in terms of hybrid dynamics and phase transitions. Theoretically, our analysis identifies 16 possible galloping footfall sequences . However, this study focuses exclusively on right-fore (FR) leading gaits due to energetic symmetries [ 12 , 13 , 14 ] . • Duty factor: d_{t}\in[0,1):=(t^{\text{LO}}_{i}-t^{\text{TD}}_{i})\mod 1 . When t^{\text{LO}}_{i}<t^{\text{TD}}_{i} , the modulo operation accounts for the periodicity of the stride cycle by effectively adding 1 before taking the difference, ensuring a consistent and normalized duty factor between 0 and 1, representing

Read the complete text