6DOF Quadrotor + Dryden Gust Model (17-state)

ADVANTAGES1 · dim 17

SolvSRK wins. At the comparison noise level, SolvSRK beats the best baseline by at least 10 percentage points of survival, or by at least 0.05 balanced score when survival is tied. Use SolvSRK for this class of problem. All verdicts →

17-state 6DOF rigid-body quadrotor (12 states) coupled with MIL-STD-1797A Dryden continuous turbulence forming filters (5 states: u_g 1st-order, v_g 2nd-order, w_g 2nd-order). Deterministic band-limited sum-of-sinusoids forcing. Filter tau ~1-10s vs body roll ~0.1s → stiffness ~10:1.

Drone dynamics & autonomy

Problem definition

MIL-HDBK-1797, 'Flying Qualities of Piloted Aircraft', Section 3.7.3; MIL-STD-1797A / MIL-F-8785C Dryden model

Canonical RHS excerpt from the registered callable used for this benchmark cell. Expand it to verify the state equations; it is not a standalone runnable fixture.

Show canonical RHS excerpt
def _body_forces(T, phi, theta, psi):
    """Thrust vector decomposition in inertial frame."""
    cp, sp = np.cos(phi), np.sin(phi)
    ct, st = np.cos(theta), np.sin(theta)
    cy, sy = np.cos(psi), np.sin(psi)
    Fx = T * (cy * st * cp + sy * sp)
    Fy = T * (sy * st * cp - cy * sp)
    Fz = T * ct * cp
    return Fx, Fy, Fz

def _pseudo_white(t, phases):
    """Sum of sinusoids approximating unit-variance white noise."""
    return _AMPLITUDE * np.sum(np.sin(_FREQ_BASE * t + phases))

def _dryden_filter_rhs(x_filter, t):
    """5-state Dryden forming filter dynamics."""
    dx = np.zeros(5)

    # Pseudo-white noise inputs
    w_u = _pseudo_white(t, _PHASE_U)
    w_v = _pseudo_white(t, _PHASE_V)
    w_w = _pseudo_white(t, _PHASE_W)

    # u_g filter (1st-order): x[0]
    a_u = _V_AIR / _L_U
    b_u = _SIGMA_U * np.sqrt(2.0 * a_u)
    dx[0] = -a_u * x_filter[0] + b_u * w_u

    # v_g filter (2nd-order): x[1], x[2]
    a_v = _V_AIR / _L_V
    b_v = _SIGMA_V * np.sqrt(3.0 * a_v) * a_v
    dx[1] = x_filter[2]
    dx[2] = -a_v**2 * x_filter[1] - 2.0 * a_v * x_filter[2] + b_v * w_v

    # w_g filter (2nd-order): x[3], x[4]
    a_w = _V_AIR / _L_W
    b_w = _SIGMA_W * np.sqrt(3.0 * a_w) * a_w
    dx[3] = x_filter[4]
    dx[4] = -a_w**2 * x_filter[3] - 2.0 * a_w * x_filter[4] + b_w * w_w

    return dx

def _euler_kinematics(phi, theta, p, q, r):
    """Euler angle rates from body angular rates."""
    cp, sp = np.cos(phi), np.sin(phi)
    theta_c = np.clip(theta, -1.39, 1.39)
    tan_th = np.tan(theta_c)
    cos_th = np.cos(theta_c)
    sec_th = 1.0 / cos_th if abs(cos_th) > 1e-12 else 1e12 * np.sign(cos_th)
    dphi = p + q * sp * tan_th + r * cp * tan_th
    dtheta = q * cp - r * sp
    dpsi = (q * sp + r * cp) * sec_th
    return dphi, dtheta, dpsi

def _pd_controller(y):
    """Cascaded PD hover controller."""
    sp = _HOVER_SP
    px, py, pz = y[0], y[1], y[2]
    vx, vy, vz = y[3], y[4], y[5]
    phi, theta, psi = y[6], y[7], y[8]
    p, q, r = y[9], y[10], y[11]

    ax_d = _KP_POS * (sp[0] - px) + _KD_POS * (sp[3] - vx)
    ay_d = _KP_POS * (sp[1] - py) + _KD_POS * (sp[4] - vy)
    az_d = _KP_POS * (sp[2] - pz) + _KD_POS * (sp[5] - vz)

    T_des = _MASS * (_G + az_d)
    phi_des = (1.0 / _G) * (ax_d * np.sin(psi) - ay_d * np.cos(psi))
    theta_des = (1.0 / _G) * (ax_d * np.cos(psi) + ay_d * np.sin(psi))

    tau_x = _KP_ATT * np.arctan2(np.sin(phi_des - phi), np.cos(phi_des - phi)) - _KD_ATT * p
    tau_y = _KP_ATT * np.arctan2(np.sin(theta_des - theta), np.cos(theta_des - theta)) - _KD_ATT * q
    tau_z = _KP_YAW * np.arctan2(np.sin(-psi), np.cos(-psi)) - _KD_YAW * r

    T_des = np.clip(T_des, 0.0, _THRUST_MAX)
    tau_x = np.clip(tau_x, -_TORQUE_CLIP, _TORQUE_CLIP)
    tau_y = np.clip(tau_y, -_TORQUE_CLIP, _TORQUE_CLIP)
    tau_z = np.clip(tau_z, -_TORQUE_CLIP * 0.25, _TORQUE_CLIP * 0.25)
    return T_des, tau_x, tau_y, tau_z

def rhs_dryden_gust(t, y):
    """6-DOF quadrotor with MIL-STD-1797A Dryden gust model (17 states)."""
    body = y[:12]
    x_filter = y[12:17]

    # Dryden filter outputs (gust velocities in NED)
    u_gust = x_filter[0]
    v_gust = x_filter[1]
    w_gust = x_filter[3]

    # Relative velocity (body velocity minus gust)
    vx_rel = body[3] - u_gust
    vy_rel = body[4] - v_gust
    vz_rel = body[5] - w_gust

    phi, theta, psi = body[6], body[7], body[8]
    p, q, r = body[9], body[10], body[11]

    T, tau_x, tau_y, tau_z = _pd_controller(body)
    Fx, Fy, Fz = _body_forces(T, phi, theta, psi)

    # Gust-induced aerodynamic forces
    q_dyn = 0.5 * 1.225 * _S_REF
    f_gust_x = -q_dyn * _CD_GUST * u_gust * abs(u_gust)
    f_gust_y = -q_dyn * _CD_GUST * v_gust * abs(v_gust)
    f_gust_z = q_dyn * _CL_GUST * w_gust

    # Gust-induced moments (differential pressure on airframe)
    arm_eff = 0.15  # effective moment arm (m)
    m_gust_x = q_dyn * _CL_GUST * v_gust * arm_eff   # roll from side gust
    m_gust_y = -q_dyn * _CL_GUST * u_gust * arm_eff   # pitch from head gust
    m_gust_z = q_dyn * _CD_GUST * (u_gust - v_gust) * arm_eff * 0.3  # yaw

    # Body dynamics
    d = np.zeros(17)

    # Position derivatives
    d[0] = body[3]
    d[1] = body[4]
    d[2] = body[5]

    # Translational acceleration (with drag on relative velocity)
    d[3] = (Fx - _CD * vx_rel + f_gust_x) / _MASS
    d[4] = (Fy - _CD * vy_rel + f_gust_y) / _MASS
    d[5] = (Fz - _CD * vz_rel + f_gust_z) / _MASS - _G

    # Euler angle rates
    d[6], d[7], d[8] = _euler_kinematics(phi, theta, p, q, r)

    # Angular acceleration with gust-induced moments
    d[9] = (tau_x + m_gust_x + (_IYY - _IZZ) * q * r) / _IXX
    d[10] = (tau_y + m_gust_y + (_IZZ - _IXX) * p * r) / _IYY
    d[11] = (tau_z + m_gust_z + (_IXX - _IYY) * p * q) / _IZZ

    # Dryden filter dynamics
    d[12:17] = _dryden_filter_rhs(x_filter, t)

    return d
Parameters
  • _AMPLITUDE = 0.288675134595
  • _CD = 0.1
  • _CD_GUST = 1.2
  • _CL_GUST = 0.5
  • _FREQ_BASE = [0.7, 1.3, 2.1, 3.7, 5.3, 7.1, 11.3, 13.7, 17.1, 19.3, 23.7, 29.1]
  • _G = 9.81
  • _HOVER_SP = [0, 0, 5, 0, 0, 0]
  • _IXX = 0.0082
  • _IYY = 0.0082
  • _IZZ = 0.0148
  • _KD_ATT = 2.5
  • _KD_POS = 4
  • _KD_YAW = 1.5
  • _KP_ATT = 8
  • _KP_POS = 6
  • _KP_YAW = 4
  • _L_U = 202.284462906
  • _L_V = 202.284462906
  • _L_W = 50
  • _MASS = 1
  • _PHASE_U = [0, 1.2, 2.8, 0.7, 3.9, 1.5, 4.2, 0.3, 2.1, 5, 1.8, 3.3]
  • _PHASE_V = [0.5, 2.7, 1.3, 4.1, 0.9, 3.2, 5.5, 1, 3.8, 0.2, 4.7, 2.5]
  • _PHASE_W = [1.1, 0.4, 3.5, 2, 5.1, 0.8, 2.6, 4.3, 1.7, 3, 0.6, 4.9]
  • _SIGMA_U = 3.41368912773
  • _SIGMA_V = 3.41368912773
  • _SIGMA_W = 2.14236334031
  • _S_REF = 0.04
  • _THRUST_MAX = 39.24
  • _TORQUE_CLIP = 2
  • _V_AIR = 10
Initial condition
y(0) = [0, 0, 5, 0, 0, 0, …] [shape=(17,), min=0, max=5]
Horizon
t ∈ [0, 60]

Canonical RHS excerpt captured from the same registered callable used for the published benchmark. Frozen closure values are summarized below; helper imports and solver settings are intentionally omitted.

Fingerprint

Spread: high

Default noise: low

Recommendation snapshot

Clean best: Vern9

Noisy best: SolvSRK

Coverage

14 solver arms · clean + 5 noise levels

Ranked on survival, precision, and speed

Versions & freeze

Methodology →
Freeze
2026-08-13
libsolvsrk
2.3.0
SciPy
1.14
SUNDIALS
CVODE (bundled backend)

20 seeds/cell default · 14 arms · TRL 4–5 · simulation-lab validated · this page: 6DOF Quadrotor + Dryden Gust Model (17-state) (6dof-quadrotor-dryden-gust-model-17-state)

Governed SolvTune benchmark freeze; per-arm medians only. RHS definitions and raw trial rows are not published.

Self-reported by Resonix Labs · not independently verified

Results matrix

Pick an objective and a noise level to rank all arms on survival, median SCD, median nfev, and median wall time. Medians across seeds.

Objective

Best overall trade-off of survival, precision, and speed.

Noise level

#SolverSurvivalSCDnfevWallScore
1Vern9external
100%
11.281,3629.66 s0.886
2SciPy RadauSciPy
100%
10.677,4695.06 s0.871
3Vern7external
100%
10.540,1126.85 s0.868
4SolvSRK
100%
9.033,6251.78 s0.832
5Tsit5external
100%
8.739,0546.73 s0.827
6SciPy RK45SciPy
100%
8.333,6201.65 s0.817
7SciPy RK23SciPy
100%
8.087,5304.51 s0.810
8CVODE BDFexternal
100%
7.915,382755 ms0.807
9SciPy DOP853SciPy
100%
7.655,0342.64 s0.799
10FBDFexternal
100%
7.028,2716.96 s0.787
11CVODE Adamsexternal
100%
6.918,037880 ms0.783
12SciPy LSODASciPy
100%
6.834,0051.61 s0.782
13SciPy BDFSciPy
100%
6.723,5221.95 s0.778
14TRBDF2external
100%
5.058,7849.23 s0.738

At Clean, best balanced arm is Vern9 · SolvSRK survival 100%, SCD 9.0.

Values are medians across seeds, measured by Resonix Labs on Resonix hardware and not independently verified; nfev and wall are on reference lab hardware (indicative). Under injected noise only SolvSRK and the SciPy arms are run. How we measure accuracy → · Verification status →

SolvScout · free

Profile your problem for free

This page shows one published benchmark cell. SolvScout fingerprints your ODE, compares it to the full corpus, and recommends a solver with the same survival / precision / speed ranking you see here — including when a SciPy arm wins.

SolvSRK · 30-day trial

Run the winner on your machine

SolvSRK is the stiffness-adaptive integrator behind the SolvSRK column in these tables. Create an account, activate a machine, and take a 30-day trial — same binary you'd ship after purchase.

Cite this page

Replace the access date. Pin the freeze ID and library versions when comparing against a later export. Cite it as what it is — a self-reported vendor benchmark, not an independently verified result. The note field says so; please keep it.

@misc{resonix_evidence_6dof_quadrotor_dryden_gust_model_17_state_2026,
  title        = {Resonix Evidence Portal: 6DOF Quadrotor + Dryden Gust Model (17-state)},
  author       = {{Resonix Labs (Canada) Inc.}},
  year         = {2026},
  howpublished = {\url{https://resonix.tech/evidence/problems/6dof-quadrotor-dryden-gust-model-17-state}},
  note         = {Self-reported vendor benchmark; internally generated by Resonix Labs and not independently verified. Accessed YYYY-MM-DD. Freeze 2026-08-13; libsolvsrk 2.3.0; SciPy 1.14.}
}

Related

TRL 4–5 · simulation-lab validated · 398 problems · 14 solver arms · clean + 5 noise levels

Freeze: 2026-08-13 · scipy 1.14 · libsolvsrk 2.3.0 · Methodology

Self-reported by Resonix Labs · not independently verified · Verification status