Hybrid Locomotion for Unstructured Environments

This direction is merged into another direction: hybrid locomotion. Its last version stays readable and citable here.

Version 1 of 144 papers44 conceptsfrozen 2026-10-04

In robotics, specialization is a double-edged sword. A robot designed for the smooth floors of a warehouse will struggle to climb a single step, while a robot designed to scale mountain faces will be agonizingly slow on a paved road. This fundamental conflict—the trade-off between efficiency on predictable terrain and versatility in unpredictable environments—has led to a new paradigm: hybrid locomotion.

Instead of choosing one mode of travel, researchers are developing systems that combine disparate mechanical strategies, such as wheels, legs, and rotors, to navigate the "unstructured" world of disaster zones, deep oceans, and planetary surfaces.

The Efficiency-Versatility Trade-off

The most common conflict in robotic design is between wheels and legs. As Atsushi Yamashita and colleagues noted in their survey of mobile mechanisms, wheels are the gold standard for energy efficiency and speed on flat surfaces. However, they are easily immobilized by rubble or steep inclines. Legged systems, by contrast, can step over obstacles and navigate ruins, but they require complex control and consume significantly more power.

To bridge this gap, researchers like Ruotian Wang and Max Schwarz have developed hybrid "wheel-leg" systems. In the Momaro robot, wheels are mounted at the ends of flexible legs. This allows the robot to "skate" on flat ground and "step" over debris. C. Dario Bellicoso takes this further by treating wheels as "torque-controlled moving footholds." By integrating wheel dynamics directly into a whole-body motion controller, the robot doesn't just switch between modes; it can walk and drive simultaneously, achieving an 83% reduction in energy consumption compared to pure walking.

Solving the Torque-Speed Mismatch

When a robot moves between different physical media—such as air and water—it encounters a "torque-speed mismatch." Aerial propellers must spin at high speeds in low-density air to generate lift, whereas aquatic propellers require high torque to push through dense water.

The TJ-FlyingFish, developed by Xuchen Liu and his team, solves this with a dual-speed gearbox. By mechanically shifting gears, the robot optimizes its motor output for each medium, allowing it to transition from a 6-minute hover in the air to a 40-minute swim underwater. Similarly, Derek Chen found that for screw-based amphibious robots, the physics of the environment dictates the strategy: in water, pitch angle drives performance, but in "quicksand-like" saturated media, the robot is more efficient if it rolls the screw like a wheel rather than spinning it.

Morphological Intelligence and Shape-Shifting

If a robot’s environment changes drastically, perhaps the robot’s body should too. This is the core of morphological intelligence. Dylan S. Shah argues that physical shape is a form of "embodied" software. A robot that can change its footprint or center of mass can resolve spatial conflicts that software alone cannot.

For example, Xinyu Zhang developed a land-air robot that uses a "staggered folding" mechanism. It maintains a wide stance for stable flight but shrinks its footprint by 79% to navigate narrow corridors on the ground. In the realm of micro-robotics, Gianluca Iaccarino used Kresling origami to create a millirobot that spins like a propeller in fluids but rolls or flips on land, using its own structural geometry to adapt to the medium.

Unified Control for Hybrid Systems

The challenge of hybrid locomotion isn't just mechanical; it's computational. How does a robot decide when to fly and when to roll? Tara Bartlett and Ali-akbar Agha-mohammadi addressed this for the "Rollocopter" by leveraging differential flatness. This mathematical property allows a single planning architecture to treat rolling and flying as different expressions of the same underlying physics, enabling the robot to consume five times less energy by rolling whenever possible.

For legged robots on treacherous terrain, Fabian Jenelten introduced Deep Tracking Control (DTC). This approach combines the precision of model-based planning (knowing exactly where to put a foot) with the robustness of reinforcement learning (knowing how to react when a foot slips). By unifying these methods, robots can navigate "stepping stone" environments with the agility of an animal and the precision of a machine.

The Actuator Tax

A recurring hurdle in this field is what Jianxiang Zhou calls the "actuator tax." Adding more modes of travel usually means adding more motors, which increases weight and reduces battery life. The SWheg project attempts to minimize this tax through "minimalist actuation," using clever linkages to switch between wheels and legs without a massive increase in hardware complexity.

By focusing on these mechanical and algorithmic "handshakes" between different modes of movement, researchers are moving toward robots that don't just operate in a specific niche, but can truly navigate the complexity of the real world.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_hybrid_locomotion_for_unstructured_envir_f7ac9707,
  title = {Hybrid Locomotion for Unstructured Environments},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/hybrid-locomotion-for-unstructured-environments/rdr_f7ac97075de940dc97576eb5afb7cb10},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_hybrid_locomotion_for_unstructured_envir_f7ac9707_v1,
  title = {Hybrid Locomotion for Unstructured Environments},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/hybrid-locomotion-for-unstructured-environments/rdr_f7ac97075de940dc97576eb5afb7cb10/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}