Researchers are increasingly focused on “pronking” – a dynamic, periodic jumping gait for quadruped robots – as a means of achieving robust and versatile locomotion, according to recent publications in the fields of robotics and computer science.
The term, coined within the robotics community, describes a specific type of explosive movement where a quadruped repeatedly leaps forward. A paper published on arXiv in April 2025 details a “dual-legged actuated spring-loaded inverted pendulum with trunk rotation” template model designed to explicitly model compliance in quadrupedal robots, enabling the generation of pronking, froggy jumping, and hop-turn motions. The research, authored by Jiatao Ding and six other researchers, aims to address the challenges of controlling robots with parallel compliance, which enhances locomotion performance but complicates the control process.
The development of this model utilizes a dual-layer trajectory optimization process, taking into account singularity-free body rotation. Hardware experiments, conducted with both rigid and newly designed compliant quadrupeds, demonstrated the template model’s ability to generate versatile dynamic motion and the benefits of parallel elasticity for explosive movement. The researchers found their model improved upon existing templates in terms of accuracy, and generalization.
Pronking isn’t limited to theoretical models. Another study, published by IEEE, focuses on increasing the robustness of pronking in articulated soft quadrupeds through a combination of model-based trajectory optimization and iterative learning control. This research employs a reduced-order soft anchor model to generate periodic forward jumping, aiming to improve performance in real-world scenarios where unmodelled dynamic effects are common.
The broader implications of this research extend beyond robotics. An article published by The Economist on February 19, 2026, notes the importance of “human equivalents of pronking” in workplace dynamics, suggesting that embracing iterative, seemingly inefficient processes can yield positive results. The article uses the example of a manager thanking a team member, highlighting the various options available and implicitly drawing a parallel to the iterative refinement seen in robotic gait control.
Archyde.com documented optimization history plots for pronking, spanning 30 minutes, as part of the research process, illustrating the iterative nature of refining robotic gait control.
