Researchers have developed a bioinspired quadruped robot with an adaptive spine and adjustable posture, enabling it to efficiently navigate complex, uneven, and confined terrains without relying on sensors.
Study: A bio-inspired adjustable posture quadruped robot with laterally undulating spine for terradynamically challenging environments. Image Credit: Syahrul Zidane As Sidiq/Shutterstock.com
Published in Nature, the study introduces a robot that actively shifts its height and width to clear obstacles and move through narrow spaces. By using undulating spinal movements, it maintains balance and stability across varied environments. In controlled tests, the robot navigated 10° slopes, tight spaces, and rough terrain with ease—achieving up to 30 % faster speeds in optimized postures—offering a low-energy, mechanically adaptive approach for scenarios like search-and-rescue and field exploration.
Background
Reptiles adapt to diverse terrains by altering their posture and using spinal undulation. While some robots mimic this flexibility or adjust posture, most don’t do both effectively. Designs that attempt to integrate these features often face trade-offs in control complexity or stability.
This study addresses that challenge by combining both posture shifting and active lateral spinal motion in a single system. The robot's symmetrical parallelogram mechanism allows it to transition from a sprawled to a semi-erect stance while keeping its center of gravity stable, enabling it to traverse slopes, tight corridors, and uneven ground without the need for sensory input.
Design and Implementation
The robot includes eight rotational axes: four for leg movement (each with one degree of freedom), two for synchronized posture adjustments, and two for spinal articulation, controlling both pitch and yaw. The posture system shifts between -60° (sprawled) and +40° (semi-erect), optimized in SolidWorks to avoid interference and maintain stability. A two-degree-of-freedom joint connects the front and rear body segments, enabling lateral undulation. Its C-shaped legs (30–35 mm diameter) support better adaptability to irregular ground.
Built from 3D-printed ABS plastic, the prototype benefits from structural flexibility, with compliant legs aiding ground interaction. Seven Dynamixel motors—four for the legs, two for posture control, and one for spine motion—are controlled via MATLAB in an open-loop system, eliminating the need for sensors.
The robot uses a lateral sequence creeping gait to maintain its center of gravity within a stable triangle during 76 % of each gait cycle. Stability held across all postures, with brief instability during leg transitions resolved quickly. Power consumption ranged from 15 to 17.7 mW on flat ground, with the 0° posture being the least efficient due to added body movement during undulation. Simulations (30–45 minutes each) validated terrain adaptability and energy efficiency before physical tests.
Locomotion Across Varied Terrains
The robot’s adaptable design allowed it to handle a range of terrains. On flat ground, -60° postures performed best on smooth surfaces thanks to sliding rear legs, while 0° and 40° postures offered better traction on rough ground, where partial forward tipping over obstacles increased stride length.
To pass through a 195 mm-wide corridor, the robot reduced its body width from 202 mm (at 0°) to 150 mm (at -60°). For vertical clearance, it adjusted its height from 240 mm (-60°) to 180 mm (40°), easily fitting through a 200 mm-high tunnel.
Obstacle clearance was posture-dependent: a 38 mm object was crossed using a 20° configuration, while a 50 mm cube required shifting to -40°. On slopes, the robot climbed a 10° incline in a 40° posture with a 60° spinal undulation amplitude (αpeak), while lower postures proved better for downhill movement. Greater undulation (αpeak up to 70° at -60°) also helped stabilize motion on 5° inclines.
Outdoor tests confirmed its real-world adaptability. The robot navigated rocky surfaces by shifting posture (from 40° to -20°) and adjusted from -60° to 40° to enter a building. Failures included leg stalls on rough ground and tipping during steep declines, showing that while sensorless adaptation works well, terrain type and incline still affect performance.
Conclusion
By integrating posture control with active spinal undulation, this study presents a quadruped robot capable of adapting to real-world terrains without relying on external sensors or AI. Its ability to shift height, width, and gait mid-movement allowed it to clear obstacles, climb slopes, and maneuver tight spaces—all through mechanical design alone.
Future developments could focus on adding compliance and sensor feedback to further enhance autonomy. For now, the robot offers a compelling approach to energy-efficient, terrain-aware mobility in settings ranging from disaster zones to field exploration.
Journal Reference
Dutta, S.K., Ozkan-Aydin, Y. A bio-inspired adjustable posture quadruped robot with laterally undulating spine for terradynamically challenging environments. Sci Rep 15, 27143 (2025). DOI:10.1038/s41598-025-07623-0. https://www.nature.com/articles/s41598-025-07623-
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.