Why Is a Moving Bicycle Stable? The Whipple Model

1. The Gyroscope Myth

A common explanation claims spinning wheels resist tipping the way a gyroscope resists reorientation. Experiments disprove this as the primary cause: bicycles built with counter-rotating extra wheels (cancelling the net angular momentum) are still self-stable, and bicycles with the front wheel's trail reversed become unstable at any speed even with normal gyroscopic wheels. Gyroscopic effects contribute a secondary steering torque, but the dominant mechanism is the geometric coupling described by the full Whipple model.

2. The Whipple Model — Four Rigid Bodies

The benchmark Whipple bicycle model treats the machine as four interconnected rigid bodies, ignoring tyre deformation, frame flex, and rider control (an unactuated, "hands-off" system):

Two degrees of freedom describe small perturbations from the upright, straight-line rolling equilibrium: lean angle φ (roll of the rear frame) and steer angle δ (rotation of the fork relative to the frame).

3. Linearized Equations of Motion

Linearizing about the upright, constant-forward-speed (v) equilibrium yields a compact second-order matrix equation:

4. Eigenmodes: Weave, Capsize, Wobble

Substituting q = q₀·e^(λt) turns the equations into a speed-dependent eigenvalue problem. Three characteristic modes emerge from typical bicycle parameters:

5. Trail, Rake, and Steering Geometry

Mechanical trail — the horizontal distance between the steering axis's ground contact point and the front wheel's contact patch — is the single most influential geometric parameter:

6. The Self-Stable Speed Range

Whipple-model studies (Meijaard, Papadopoulos, Ruina, Schwab, 2007) confirmed with a physical "uncontrolled" bicycle that self-stability is real and quantitatively matches the linearized eigenvalue predictions:

• Below v_weave (~4 m/s): the weave mode is unstable — the classic wobble a beginner feels at walking pace, requiring active steering correction.
• Between v_weave and v_capsize: both weave and capsize modes decay — the bicycle is self-stable and will recover from small perturbations hands-off.
• Above v_capsize (if it exists): the capsize mode becomes unstable again, though very slowly (long time constant), so it is rarely noticeable at ordinary cycling speeds.

7. JavaScript Linear Stability Check

// Evaluate Whipple-model eigenvalues at a given speed (toy matrices)
// M, C1, K0, K2 are 2x2 matrices from bicycle benchmark parameters
function stateMatrix(M, C1, K0, K2, v, g = 9.81) {
  // Build 4x4 state matrix A for x = [phi, delta, phiDot, deltaDot]
  const Minv = invert2x2(M);
  const K = addScaled(scale2x2(K0, g), K2, v * v); // g*K0 + v^2*K2
  const C = scale2x2(C1, v);
  const negMinvK = scale2x2(mul2x2(Minv, K), -1);
  const negMinvC = scale2x2(mul2x2(Minv, C), -1);
  // A = [[0, I], [-Minv*K, -Minv*C]]  (4x4 block form)
  return assembleBlock(negMinvK, negMinvC);
}

function isStable(A) {
  const eigenvalues = eig4(A); // numeric eigenvalue solver
  return eigenvalues.every(lambda => lambda.re < 0);
}

// Sweep forward speed to find the self-stable window
function findStableRange(M, C1, K0, K2, vMin = 0, vMax = 10, step = 0.05) {
  const stableSpeeds = [];
  for (let v = vMin; v <= vMax; v += step) {
    const A = stateMatrix(M, C1, K0, K2, v);
    if (isStable(A)) stableSpeeds.push(v);
  }
  return { min: stableSpeeds[0], max: stableSpeeds[stableSpeeds.length-1] };
}

8. Engineering Applications