AUV 6DOF station-keeping (Fossen)

ADVANTAGES1 · dim 12

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 →

12D Fossen 6DOF AUV rigid-body dynamics with hydrodynamic forces - station-keeping maneuver

Defense autonomy

Problem definition

Fossen (2011) Handbook of Marine Craft Hydrodynamics

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 _angular_rate_transform(phi, theta):
    """Transform body angular rates [p,q,r] to Euler angle rates [dphi,dtheta,dpsi]."""
    cp, sp = np.cos(phi), np.sin(phi)
    ct = np.cos(theta)
    tt = np.tan(np.clip(theta, -1.4, 1.4))
    sec_t = 1.0 / ct if abs(ct) > 1e-12 else 1e12 * np.sign(ct)

    T = np.array([
        [1.0, sp * tt,  cp * tt],
        [0.0, cp,       -sp],
        [0.0, sp * sec_t, cp * sec_t],
    ])
    return T

def _rotation_matrix(phi, theta, psi):
    """Standard ZYX rotation matrix for NED 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)

    R = np.array([
        [cy * ct, cy * st * sp - sy * cp, cy * st * cp + sy * sp],
        [sy * ct, sy * st * sp + cy * cp, sy * st * cp - cy * sp],
        [-st,     ct * sp,                 ct * cp],
    ])
    return R

def _auv_rhs(t, y):
    """Fossen 6DOF AUV: d_eta/dt = J(eta)*nu, M*d_nu/dt = -C*nu - D*nu - g + tau."""
    eta = y[:6]   # [x, y, z, phi, theta, psi]
    nu = y[6:12]  # [u, v, w, p, q, r]

    phi, theta, psi = eta[3], eta[4], eta[5]
    u, v, w, p, q, r = nu[0], nu[1], nu[2], nu[3], nu[4], nu[5]

    # -- kinematics: d_eta/dt = J(eta) * nu --
    R = _rotation_matrix(phi, theta, psi)
    T_ang = _angular_rate_transform(phi, theta)

    d_pos = R @ nu[:3]
    d_ang = T_ang @ nu[3:]

    # -- Coriolis + centripetal (rigid-body + added-mass cross terms) --
    # Simplified: dominant cross-coupling terms
    c_u = -(_M - _YV_DOT) * v * r + (_M - _ZW_DOT) * w * q
    c_v = (_M - _XU_DOT) * u * r - (_M - _ZW_DOT) * w * p
    c_w = -(_M - _XU_DOT) * u * q + (_M - _YV_DOT) * v * p
    c_p = (_I_R - _I_Q) * q * r
    c_q = (_I_P - _I_R) * p * r
    c_r = (_I_Q - _I_P) * p * q

    # -- damping: linear + quadratic --
    d_u = _XU * u + _XUU * abs(u) * u
    d_v = _YV * v + _YVV * abs(v) * v
    d_w = _ZW * w + _ZWW * abs(w) * w
    d_p = _KP * p
    d_q = _MQ * q
    d_r = _NR * r

    # -- restoring forces (neutrally buoyant, BG offset) --
    cp_r, sp_r = np.cos(phi), np.sin(phi)
    ct_r, st_r = np.cos(theta), np.sin(theta)
    g_vec = np.array([
        -(_W - _B) * st_r,
        (_W - _B) * ct_r * sp_r,
        (_W - _B) * ct_r * cp_r,
        _BG * _B * ct_r * sp_r,
        _BG * _B * st_r,
        0.0,
    ])

    # -- acceleration: M * d_nu/dt = tau + f_hydro - C*nu - g --
    # d_* variables use SNAME sign convention (already negative for drag),
    # so they are ADDED as forces, not subtracted.
    du_dt = (_TAU[0] + d_u - c_u - g_vec[0]) / _M_U
    dv_dt = (_TAU[1] + d_v - c_v - g_vec[1]) / _M_V
    dw_dt = (_TAU[2] + d_w - c_w - g_vec[2]) / _M_W
    dp_dt = (_TAU[3] + d_p - c_p - g_vec[3]) / _I_P
    dq_dt = (_TAU[4] + d_q - c_q - g_vec[4]) / _I_Q
    dr_dt = (_TAU[5] + d_r - c_r - g_vec[5]) / _I_R

    dy = np.empty(12)
    dy[0:3] = d_pos
    dy[3:6] = d_ang
    dy[6] = du_dt
    dy[7] = dv_dt
    dy[8] = dw_dt
    dy[9] = dp_dt
    dy[10] = dq_dt
    dy[11] = dr_dt
    return dy
Parameters
  • _B = 490.5
  • _BG = 0.02
  • _I_P = 1.5
  • _I_Q = 12
  • _I_R = 7
  • _KP = -5
  • _M = 50
  • _MQ = -10
  • _M_U = 55
  • _M_V = 70
  • _M_W = 80
  • _NR = -5
  • _TAU = [2, 0, 0, 0, 0, 0]
  • _W = 490.5
  • _XU = -10
  • _XUU = -20
  • _XU_DOT = -5
  • _YV = -20
  • _YVV = -40
  • _YV_DOT = -20
  • _ZW = -30
  • _ZWW = -60
  • _ZW_DOT = -30
Initial condition
y(0) = [0, 0, -10, 0, 0, 0, 0.1, 0, 0, 0, 0, 0]
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: low

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: AUV 6DOF station-keeping (Fossen) (auv-6dof-station-keeping-fossen)

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%
11.92904.04 s0.902
2Vern7external
100%
11.62423.95 s0.895
3SolvSRK
100%
10.945910 ms0.878
4SciPy DOP853SciPy
100%
10.51584 ms0.869
5SciPy RadauSciPy
100%
10.344118 ms0.863
6SciPy RK45SciPy
100%
9.01945 ms0.834
7Tsit5external
100%
9.0174668 ms0.833
8SciPy RK23SciPy
100%
9.042211 ms0.832
9SciPy LSODASciPy
100%
8.71513 ms0.827
10FBDFexternal
100%
8.01585.12 s0.809
11CVODE Adamsexternal
100%
7.811111 ms0.806
12SciPy BDFSciPy
100%
7.821412 ms0.804
13CVODE BDFexternal
100%
7.513811 ms0.797
14TRBDF2external
100%
6.85905.06 s0.782

At Clean, best balanced arm is Vern9 · SolvSRK survival 100%, SCD 10.9.

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_auv_6dof_station_keeping_fossen_2026,
  title        = {Resonix Evidence Portal: AUV 6DOF station-keeping (Fossen)},
  author       = {{Resonix Labs (Canada) Inc.}},
  year         = {2026},
  howpublished = {\url{https://resonix.tech/evidence/problems/auv-6dof-station-keeping-fossen}},
  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