💻 computer science

Hip Energized Monopedal Hopping

This paper introduces a novel hip-energized monopedal hopping strategy that utilizes pitch stabilization torques to counteract damping losses and regulate gait energy, a method validated through hybrid averaging analysis, simulations, and physical experiments on the Penn Jerboa robot achieving stable speeds up to 8.85 leg lengths per second.

Original authors: Shane Rozen-Levy, Griffon McMahon, Daniel Koditschek

Published 2026-08-12

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Shane Rozen-Levy, Griffon McMahon, Daniel Koditschek

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where robots don't just shuffle along on stiff, pre-programmed tracks, but actually run with the same springy, bouncing energy as a kangaroo or a human. This is the exciting corner of robotics known as legged locomotion. For decades, scientists have been trying to teach machines to hop and run by mimicking a simple idea called the "Spring-Loaded Inverted Pendulum" (SLIP). Think of this like a pogo stick: the robot's leg acts as a spring that compresses when it hits the ground and bounces it back up. The tricky part is that real life isn't perfect; air resistance and friction steal energy away with every bounce, so the robot would eventually stop unless someone keeps pumping energy back in. Usually, robots do this by using powerful motors in their legs to push against the ground. But what if we could make a robot run faster and more efficiently by using a different trick entirely, one that turns the robot's own balance into a source of power? That is the big question this paper tackles.

The researchers behind this study, working with a robot named the "Penn Jerboa," discovered a clever new way to make a one-legged robot (a monoped) hop faster and more stably. Instead of just using motors to push the leg down, they figured out how to use the robot's hip motor to do double duty: keeping the robot from falling over and adding energy to the bounce at the same time. It's like a gymnast on a trampoline who leans their body forward just right; that lean creates a force that not only keeps them upright but also propels them higher and faster with every jump.

The paper introduces a "hip-energized" control strategy. In simple terms, the robot's brain (its controller) constantly adjusts the robot's center of mass—shifting it slightly forward or backward—while the robot is in the air. This shift creates a torque (a twisting force) at the hip. Usually, engineers use hip motors just to stop the robot from tipping over. But this team realized that if they carefully manage how the robot leans, that same twisting force can actually push the robot forward and upward, fighting against the energy loss from friction.

To make this work, the robot also needs to change where it puts its foot down. The paper proves that for this energy-boosting trick to work, the robot cannot land with its foot in a perfectly symmetric spot. It has to land slightly differently than it took off, creating an "asymmetric" gait. Think of it like a runner who doesn't just step straight down but leans into their stride; this slight imbalance is actually necessary to keep the energy flowing. The researchers used advanced math (called "hybrid averaging") to predict exactly how the robot should move, and they built a controller that tells the robot how to lean and where to step.

The results are impressive. In computer simulations and real-world tests on the Penn Jerboa, this new strategy allowed the robot to hop at speeds ranging from 1.02 m/s to 1.77 m/s. To put that in perspective, that's about 5.1 to 8.85 leg lengths per second. This is significantly faster than previous attempts on the same robot, which only managed speeds up to 1.0 m/s. The paper shows that by using the hip to stabilize pitch (the up-and-down tilt of the body) and simultaneously energize the bounce, the robot can run much more efficiently. The math behind it is solid, with the researchers providing formal proofs that the robot will stay stable under these conditions, and the physical experiments confirm that the robot can indeed hop endlessly at these higher speeds without falling over.

However, the paper is careful to note that this isn't a magic bullet for every robot. The strategy relies on specific assumptions, like the robot having a way to shift its center of mass (which the Jerboa does with its tail). The researchers also point out that while their model predicts the robot's behavior quite well, the real robot sometimes moves a bit faster than the math predicts because of the unique way its legs are built. But overall, the study suggests that by "leaning in" and using hip torque creatively, we can make legged robots much more agile and energetic, potentially leading to machines that can run faster and handle rougher terrain in the future.

Technical Summary: Hip Energized Monopedal Hopping

Problem Statement
Recent advancements in legged robotics have largely converged on highly actuated limbs (e.g., 3 motors per leg) that allow for arbitrary ground reaction force planning. However, this paradigm raises questions regarding power efficiency and the necessity of such high actuation density, particularly given mass-specific power limitations. The authors investigate whether a biped or monoped with fewer actuators can achieve agile, high-speed locomotion by better utilizing available power. Specifically, the paper addresses the challenge of controlling a pitch-unlocked planar monoped (a 3-DoF Spring-Loaded Inverted Pendulum or SLIP with attitude) using only a single hip actuator. Previous work on the Penn Jerboa robot utilized a tail to energize the system, but this approach relied on the tail acting as a primary energy source. This paper explores a "hip-energized" strategy where the hip torque is used not only to stabilize body pitch but also to inject energy into the translational subsystem to counteract damping losses, potentially enabling higher speeds and more robust hopping.

Methodology
The authors propose a novel control strategy and a corresponding analytical framework:

Control Strategy:
- Hip Torque Control: A continuous-time PD controller with a discrete-time feedforward integrator stabilizes the body pitch. Crucially, the feedforward term ( $\bar{\tau}$ ) is adjusted based on pitch error to counteract the disturbance caused by the center of mass (CoM) offset. This same torque term energizes the translational subsystem.
- Stepping Controller: A discrete-event stepping policy adjusts the leg touchdown angle. Unlike symmetric strategies (e.g., Raibert's neutral point), this controller employs an asymmetric stepping strategy. It redistributes energy between the radial (leg extension) and angular (leg swing) degrees of freedom to counteract dissipative losses and maintain a user-specified balance between forward speed and apex height.
- Implementation Enhancements: To improve the basin of attraction for physical experiments (where initial conditions are variable), the authors implement a discrete energy loop that adjusts the CoM offset ( $d_x$ ) between strides and a feedforward term for the hip torque bias.
Analytical Framework (Hybrid Averaging):
- The authors decompose the 3-DoF system into a cascaded structure: an isolated 1-DoF pitch subsystem that perturbs a 2-DoF translational SLIP subsystem.
- They apply hybrid averaging (extending the work of De et al., 2018) to the full 2-DoF SLIP dynamics. This requires a novel set of coordinates ( $z$ ) that treat the SLIP as a unitary 2-DoF system rather than a parallel composition of 1-DoF radial and angular systems.
- The analysis yields closed-form expressions for the fixed points and eigenvalues of the resulting gait, providing formal stability guarantees (Theorem 1) and revealing the necessity of stepping asymmetry for hip-energized gaits (Theorem 2).

Key Contributions

Novel Control Strategy: A hip-only actuated controller for a 3-DoF pitch-unlocked SLIP that stabilizes pitch while simultaneously energizing the translational motion.
Formal Stability Analysis: The first application of hybrid averaging to a physically realistic, full 2-DoF SLIP model with attitude. This provides explicit functional expressions for fixed points and stability conditions, moving beyond numerical simulation or linearized approximations.
The Necessity of Asymmetry: Theoretical proof (Theorem 2) that a hip-energized gait requires an asymmetric stepping strategy to prevent unbounded growth in angular energy while radial energy decays.
Hardware Validation: Successful implementation on the Penn Jerboa robot, achieving stable hopping speeds ranging from 1.02 m/s to 1.77 m/s (5.1 to 8.85 leg lengths/s).

Results

Simulation: Simulations on a generic 5-link biped and the Penn Jerboa model showed that the analytical fixed points predicted the steady-state behavior with approximately 6–14% accuracy for fore-aft speed and height. The controller successfully regulated energy and energy ratios across a wide operating regime.
Hardware: Experiments on the physical Penn Jerboa demonstrated stable, continuous hopping. The robot achieved speeds up to 1.77 m/s, a significant improvement over previous tail-energized hopping results on the same platform (0.2–1.0 m/s).
Model Accuracy: The analytical model predicted hardware fixed points with a mean percent error of 16.4% for speed and 12.1% for apex height. Discrepancies were attributed to hardware-specific factors, such as the non-axial forces generated by the robot's four-bar leg linkages and boom disturbances, which were not fully captured in the simplified SLIP model.
Virtual Pivot Point (VPP): Although the controller does not explicitly target a VPP, the resulting ground reaction forces in simulation converged to a VPP, suggesting that VPP formation may be a consequence rather than a prerequisite of stable hip-energized gaits.

Significance and Claims
The paper claims to advance the project of developing SLIP template control schemes for generic anchoring bodies. By demonstrating that a hip-actuated monoped can achieve full dynamic sagittal plane mobility using only one motor, the work suggests that high-power hip torques can be effectively recruited to energize the system. The authors posit that their analytical insights—specifically the link between posture, hip torque, and gait asymmetry—offer a pathway for designing controllers that utilize available power more effectively than current highly actuated paradigms. The successful transition from theory to hardware on the Jerboa, achieving high speeds and stability, serves as empirical evidence that hip-energized strategies can outperform tail-energized alternatives in terms of speed and endurance, provided the control strategy accounts for the necessary asymmetry in the gait. The work concludes that while the current analysis is limited to the sagittal plane, the insights regarding the interplay of physical and control parameters are applicable to broader legged machine design.

More like this