Passive-Dynamic Walker Stability Mapping

ISEF Category: Robotics and Intelligent Machines

Subcategory: Biomechanics  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A tiny robot can walk downhill with no motor at all. That sounds like magic, but it is really physics doing the work. Small changes in foot flexibility or hip mass can decide whether the walker glides smoothly or crashes on the next step. Your project can measure exactly where that line sits.

What Is It?

A passive-dynamic bipedal walker is a robot that walks because gravity pulls it down a slope. Each step acts like a pendulum swing. The shape of the legs, the bend in the toe, and the location of the mass all affect whether the next step lands cleanly.

Think of it like a swing set with two seats attached. If the timing is right, each swing helps the next one. If the timing drifts, the motion falls apart. In robotics, a Poincaré section is a way to check the state of the system at the same point in each stride, so you can see whether the walker returns to a stable pattern or wanders away. MuJoCo is a physics simulator that can model the same motion on a computer, so you can compare real behavior to a digital version.

Why This Is a Good Topic

This is a strong science fair topic because you can change one design variable at a time and measure a clear result, like stable walking, step count, or recovery after a disturbance. It connects to real robotics and prosthetics research, where efficiency and balance matter. You can learn mechanics, modeling, simulation, and data analysis in one project, and the results are easy to graph and compare.

Research Questions

• How does toe compliance affect the size of the stability basin for a passive-dynamic bipedal walker? ?
• How does hip-pendulum mass change the number of stable downhill steps before a fall? ?
• What is the effect of ramp angle on stride repeatability for the same walker design? ?
• To what extent does changing foot geometry alter recovery after a small push or release disturbance? ?
• Which combination of toe compliance and hip mass gives the largest region of stable gait in experiment? ?
• How closely does a MuJoCo model predict the walker’s observed step-to-step return map? ?

Basic Materials

• PLA filament or 3D printed walker parts.
• Ball bearings sized for the axle design.
• Access to a 3D printer.
• Adjustable tilted ramp with a repeatable surface.
• Digital scale with 0.1 g accuracy.
• Ruler or digital calipers.
• Smartphone or high-speed camera for video capture.
• Tape, clamps, and basic hand tools.
• Marker or tracking stickers for motion analysis.
• Notebook or spreadsheet for logging trials.

Advanced Materials

• Precision 3D printed walker frames.
• Assorted bearings, axles, and spacers.
• Motion capture markers or fiducial stickers.
• Force plate or pressure-sensitive walkway.
• High-speed camera with tripod.
• Adjustable incline stage with angle readout.
• Calibrated masses for hip tuning.
• Computer with MuJoCo installed.
• Python environment for trajectory and return-map analysis.
• ImageJ for frame-by-frame kinematic tracking.

Software & Tools

• MuJoCo: Simulates the walker so you can compare real gait data with a physics model.
• Python: Organizes trial data, computes stability metrics, and plots return maps.
• ImageJ: Tracks body position frame by frame from video.
• GeoGebra: Helps you sketch geometry changes and compare leg angles.
• Google Sheets: Stores trial results and makes quick charts during testing.

Experiment Steps

1. Define the exact stability metric you will measure, such as step count, return map spread, or fall threshold.
2. Choose one design variable to change first, then hold every other part of the walker fixed.
3. Plan a repeatable trial setup with a consistent ramp, start position, and recording method.
4. Build a measurement workflow that turns video into stride data and a Poincaré section.
5. Create a simulation version of the same walker in MuJoCo and match its parameters to your physical build.
6. Compare experimental and simulated stability basins, then test whether the two models agree across several design settings.

Common Pitfalls

• Changing more than one geometry variable at once, which makes it impossible to tell whether toe compliance or hip mass caused the result.
• Using a ramp surface that shifts between trials, which adds random friction changes to the gait data.
• Measuring only whether the walker falls, which misses useful differences in stride stability before failure.
• Building a simulation that does not match the real mass distribution, which creates false agreement with the physical walker.
• Tracking the walker from a bad camera angle, which distorts step length, toe angle, and return-map measurements.

What Makes This Competitive

A stronger project will do more than show that the walker can walk. You can map a real stability basin, compare it to simulation, and explain where the model succeeds or fails. Strong entries also test uncertainty, repeat trials, and use a clear metric for gait stability instead of a vague success score. That combination makes the work feel like real biomechanics research, not just a cool demo.

Project Variations

• Test how different ramp surface materials change gait stability through friction.
• Compare a rigid toe to a layered flexible toe and measure how each affects recovery from perturbations.
• Replace the hip pendulum mass with different shapes or mass distributions and see whether inertia placement matters more than total mass.

Learn More

• MIT OpenCourseWare, Underactuated Robotics: course material on passive walking, dynamics, and stability, found by searching the MIT OpenCourseWare site.
• PubMed: search for review articles on passive dynamic walking, bipedal locomotion, and gait stability.
• NASA NTRS: search for robotics papers on legged locomotion modeling and control.
• Biomechanics and Motor Control of Human Movement: a free library or school library copy can help with gait basics and pendulum analogies.
• MuJoCo Documentation: official model-building and simulation reference, available through the MuJoCo site and docs page.