Genesis-Embodied-AI/genesis-world · error · ValueError

direction must have shape (3,)

Error message

direction must have shape (3,)

What it means

A directional ForceField (e.g. ConstantForceField) validates its direction argument at construction: after np.asarray, it must have shape exactly (3,). Any other shape (wrong length, nested list, scalar, or a (3,1) array) raises this ValueError before the field is created.

Source

Thrown at genesis/engine/force_fields.py:66

class Constant(ForceField):
    """
    Constant force field with a static acceleration everywhere.

    Parameters:
    -----------
    direction: array_like, shape=(3,)
        The direction of the force (acceleration). Will be normalized.
    strength: float
        The strength of the force (acceleration).
    """

    def __init__(self, direction=(1, 0, 0), strength=1.0):
        super().__init__()

        direction = np.array(direction)
        if direction.shape != (3,):
            raise ValueError("direction must have shape (3,)")

        self._direction = direction / np.linalg.norm(direction)
        self._strength = strength
        self._acc_qd = qd.Vector(self._direction * self._strength, dt=gs.qd_float)

    @qd.func
    def _get_acc(self, pos, vel, t, i):
        return self._acc_qd

    @property
    def direction(self):
        return self._direction

    @property
    def strength(self):
        return self._strength

View on GitHub (pinned to 56e4aa5d82)

Solutions

1. Pass a flat 3-element iterable, e.g. direction=[0,0,-1] or np.array([0,0,-1])
2. If the value comes from another array, flatten to shape (3,) first: np.ravel(arr) or arr.reshape(3)
3. Validate direction.shape == (3,) in your scene-construction code before creating the field

Example fix

# before
field = ConstantForceField(direction=[[0], [0], [-1]], strength=1.0)
# after
field = ConstantForceField(direction=[0, 0, -1], strength=1.0)

direction = np.asarray(direction).ravel()
assert direction.shape == (3,), 'direction must be a flat 3-vector'
field = ConstantForceField(direction=direction, strength=1.0)

def is_vec3(x) -> bool:
    return np.asarray(x).shape == (3,)

try:
    field = ConstantForceField(direction=direction, strength=1.0)
except ValueError as e:
    raise ValueError(f'bad direction {direction!r}: {e}') from e

When it happens

Trigger: Passing direction=[1,0] (2 elements), direction=[1,0,0,0] (4 elements), a scalar, or a (3,1)/(1,3) nested array to ConstantForceField(direction=...).

Common situations: Configuring wind-like force fields in a scene and passing a 2D vector from a config file, reusing a column-vector numpy array shaped (3,1), or passing a batch of directions where only one is expected.

Related errors

• center must have shape (3,)
• position must have shape (3,)
• probe_arc_spacing must be positive, got {arc_spacing}
• Python module 'uipc' is required by IPCCoupler but is not in
• `morph` in hybrid entity should be either URDF or Mesh

AI-assisted analysis of Genesis-Embodied-AI/genesis-world@56e4aa5d82 (2026-08-28). Data as JSON: /api/errors/a04b46bbe1d107e5. Report an issue: GitHub.