Introduction
Soft robotics offers a promising path for navigating complex, unstructured environments thanks to the inherent compliance and dexterity of soft materials. However, the field faces significant challenges in fabrication complexity and control unpredictability. Traditional soft robots are often monolithic; a single failure can render the entire system useless. Moreover, the infinite degrees of freedom in soft structures typically lead to "trial-and-error" gait development. This research addresses these hurdles by introducing a modular, hybrid soft-stiff quadrupedal robot governed by a robust kinematic framework.
Modular Design and Construction
The core innovation lies in a modular approach where the robot is assembled from five identical pneumatic modules: four serving as limbs and one as the central body. This design significantly simplifies fabrication and maintenance, as individual modules can be replaced if damaged.
Each module utilizes a hybrid construction to balance flexibility with structural integrity. As shown in the figure above, a module consists of a 3D-printed "hip" joint, an inextensible yet flexible plastic backbone (an Igus TRS-40 "dresspack" cable guide), and three pneumatic muscle actuators (PMAs). This "soft + stiff" architecture prevents buckling under load while allowing for controlled, omnidirectional bending.
System Architecture
The quadrupedal configuration leverages the modularity to create a high-degree-of-freedom system. By using an active soft module for the body rather than a rigid torso, the robot achieves a total of 15 Degrees of Freedom (DoF).
The active body module (labeled 'B' in the figure) is a critical component for maneuverability. It allows the robot to change the orientation of its "hips" dynamically, facilitating complex turning maneuvers that are difficult for rigid-bodied quadrupeds.
Kinematic Modeling and Gait Generation
To move beyond empirical gait generation, the researchers developed a comprehensive floating-base kinematic model. By treating the bending of each module as a circular arc, they established a mathematical mapping from joint-space variables (PMA length changes) to task-space coordinates (limb tip positions).
This framework enabled the parameterization of several distinct locomotion modes:
- Straight Locomotion: Achieved through phase-shifted actuation of diagonal limb pairs.
- In-place Turning: Executed by reversing the rotation of limbs on one side of the body.
- Differential Strides: Mimicking a car's differential by using different stride radii for inner and outer limbs.
- Combined Turning: A hybrid approach integrating body bending with differential strides to minimize ground slipping and maximize efficiency.
Key Results and Insights
Experimental validation using a Polhemus G4 motion tracking system demonstrated the model's high predictive accuracy. Straight locomotion speeds were predicted with an error of less than 5%, while the more complex "Combined Turning" method showed an error of approximately 9%.
A significant finding was the impact of actuation frequency on efficiency. The robot performed optimally at roughly 0.75 Hz. At higher frequencies (e.g., 1.0 Hz), the physical response time of the soft materials and air pressure became a bottleneck, preventing the limbs from completing their full intended range of motion. To account for these real-world losses, the team introduced a "locomotion efficiency" parameter ($\eta$), which allowed for accurate speed predictions across various gait parameters.
Conclusion and Future Work
The modular, hybrid design presented in this work offers a scalable and predictable platform for soft quadrupedal research. While the current prototype is tethered to an external air supply, future iterations aim to integrate on-board power and sensing. The transition from open-loop to closed-loop control using soft sensors will be essential for navigating truly unpredictable terrain where real-time feedback is required to compensate for varying surface friction and environmental obstacles.
Reference: Arachchige, D. D. K., et al. "Soft Steps: Exploring Quadrupedal Locomotion With Modular Soft Robots." IEEE Access, 2023.
Summary updated