Introduction
Designing robots capable of both flight and ground locomotion presents a significant engineering challenge. Most hybrid "aerial-ground" vehicles suffer from two primary drawbacks: the weight penalty of carrying separate drivetrains for each mode and a lack of maneuverability in confined spaces. Furthermore, robots that use propellers to push themselves on passive wheels often struggle on slippery surfaces, as they rely on ground friction to generate the centripetal force required for turning.
In this paper, we present Skater, a novel bi-modal bi-copter designed to overcome these limitations. By utilizing a longitudinally arranged bi-copter system with passive wheels, Skater achieves a lightweight, narrow profile and the ability to perform active steering on low-friction surfaces using vectored thrust.
System Design and Unified Actuation
The core innovation of Skater lies in its mechanical configuration. Unlike traditional quadrotor hybrids, Skater employs a longitudinally arranged bi-copter—where the two rotors are aligned along the direction of travel—flanked by two large passive wheels. This arrangement provides several distinct advantages:
- Unified Actuation: The same two rotors and tilting servo motors provide thrust for both flight and ground propulsion. This eliminates the need for heavy dedicated wheel motors, significantly extending potential operational time.
- Minimal Traversing Width: The longitudinal orientation makes Skater exceptionally narrow, allowing it to navigate through gaps that would be impassable for wider quadrotor-based systems.
- Vectored Thrust Steering: By tilting its rotors, the robot can generate lateral aerodynamic forces. This allows Skater to "push" itself into a curve, creating the necessary centripetal force independently of wheel-to-ground friction. This enables stable, high-speed maneuvers even on surfaces like ice or wet tile.
Modeling and Control Framework
To manage the complex transitions between aerial and ground modes, the researchers developed a comprehensive dynamic model of the system. A key finding was the system's differential flatness in ground mode, a mathematical property that simplifies trajectory planning and control.
Unlike many bi-copters that prioritize "pendulum stability" by placing the center of mass below the rotors, Skater’s center of mass is positioned above the servos. This design choice results in a "minimum-phase system," which allows for more aggressive and precise attitude control during high-speed maneuvers.
The system is governed by a Nonlinear Model Predictive Control (NMPC) framework. This unified controller handles the robot's dynamics holistically, ensuring smooth transitions from driving to flying while respecting the physical constraints of the motors and servos.
Experimental Results
The performance of Skater was validated through a series of real-world experiments and benchmarks against state-of-the-art quadrotor hybrids.
High-Speed Locomotion and Efficiency
In ground mode, Skater achieved a maximum speed of 2.9 m/s and an acceleration of 3.0 m/s². Crucially, ground locomotion was found to be approximately 7 times more energy-efficient than flight, saving 85.8% of the power required to remain airborne. This efficiency makes the ground mode an ideal choice for long-range traversal, reserving flight for overcoming obstacles.
Slippery Surface Performance
In head-to-head tests on slippery surfaces, traditional passive-wheeled robots often slide out of control during turns. Skater, leveraging its vectored thrust, maintained a precise path with a Root Mean Square Error (RMSE) of only 0.107 m, whereas the benchmark quadrotor-based model failed to follow the trajectory.
Narrow Space Navigation
To demonstrate its compact design, Skater successfully navigated a gap only 210 mm wide. This capability highlights its utility in search-and-rescue or industrial inspection scenarios where robots must squeeze through tight corridors or debris.
Conclusion and Future Work
Skater represents a significant step forward in hybrid robot design, offering a lightweight, narrow, and highly maneuverable solution for diverse environments. By decoupling steering force from ground friction, it opens new possibilities for high-speed robotics on challenging terrains.
Future work will focus on adapting the control laws for uneven or "off-road" 3D terrain and integrating onboard sensing to transition from motion-capture-dependent testing to fully autonomous navigation in the wild.
Summary updated