# Moro 0.5.0 Release Notes

Moro 0.5.0 expands the library from symbolic robot modeling and position inverse kinematics into a broader teaching-oriented workflow for differential kinematics, local capability analysis, full-pose IK, trajectories, numerical dynamics, simulation, and sampled workspace analysis.

## Highlights

### Orientation and transformations

0.5.0 completes the accepted orientation-tooling foundation with:

- proper-Euler and Tait-Bryan sequence support;
- quaternion conversions using scalar-first ordering;
- rotation-vector conversions;
- `vex()`;
- descriptive `is_rotation_matrix()` and `is_homogeneous_transform()` predicates.

The legacy predicates `is_SO3()`, `is_SE3()`, `isrot()`, and `ishtm()` remain temporarily available and emit `DeprecationWarning`.

### Differential kinematics and local capability

The new `moro.differential_kinematics` module provides:

- task Jacobian selection;
- Cartesian velocity propagation;
- pseudoinverse and damped-least-squares velocity IK;
- singular values;
- numerical rank;
- condition number;
- singularity detection;
- Yoshikawa manipulability.

The canonical geometric-twist ordering is:

```text
(vx, vy, vz, wx, wy, wz)
```

### Full-pose inverse kinematics

`solve_pose_ik()` extends numerical IK to homogeneous-transform targets while preserving the existing position-only IK APIs.

CCD remains intentionally position-only in 0.5.0.

### Trajectory generation

The new `moro.trajectory` module provides linear, cubic, and quintic point-to-point trajectories in:

- joint space;
- Cartesian position space.

Time-series arrays follow the time-major convention:

```text
(N, dof)
```

and Cartesian position arrays use:

```text
(N, 3)
```

including planar cases.

### Numerical dynamics and simulation

The new `moro.dynamics` module provides:

- `inverse_dynamics()`;
- `forward_dynamics()`;
- `state_derivative()`;
- `simulate()`;
- `DynamicsSolution`.

Simulation uses SciPy `solve_ivp` and supports zero, constant, time-varying, and state-feedback generalized-force inputs.

### Workspace sampling

The new `moro.workspace` module provides reproducible sampled reachable-workspace analysis with:

- finite joint-space domains;
- local NumPy RNG seeding;
- symbolic parameter substitution;
- preserved configuration/point correspondence;
- Matplotlib point-cloud visualization through `plot_workspace()`.

Workspace samples are uniform in joint coordinates, not Cartesian space.

## Important migration: `dynamic_model()`

Moro 0.5.0 intentionally changes the meaning of:

```python
robot.dynamic_model()
```

In 0.4.x it returned per-joint Euler-Lagrange equations.

In 0.5.0 it returns the standard matrix manipulator equation:

```text
M(q) qdd + C(q, qd) qd + G(q) = tau
```

Code that relied on the old behavior should migrate to:

```python
robot.euler_lagrange_equations()
```

The old matrix-form helper remains temporarily available as:

```python
robot.dynamic_model_matrix_form()
```

but is deprecated and will emit `DeprecationWarning`.

## Transformation predicate migration

Prefer:

```python
from moro.transformations import (
    is_rotation_matrix,
    is_homogeneous_transform,
)
```

over the legacy utility names:

```text
is_SO3()
is_SE3()
isrot()
ishtm()
```

The legacy wrappers preserve their Boolean return contract during the deprecation period.

## Python and dependencies

Moro 0.5.x supports Python 3.11 and newer.

The main runtime dependencies are:

- SymPy;
- NumPy;
- Matplotlib;
- SciPy.

SciPy is now required because numerical time integration is a core 0.5.0 capability.

## Package API policy

New 0.5.0 capabilities are primarily module-oriented. Existing root imports remain intentionally compact.

Recommended imports include:

```python
from moro.differential_kinematics import solve_velocity_ik
from moro.inverse_kinematics import solve_pose_ik
from moro.trajectory import joint_trajectory
from moro.dynamics import simulate
from moro.workspace import sample_workspace
from moro.visualization import plot_workspace
```

## Scope intentionally deferred

Moro 0.5.0 does not add general motion planning, collision-aware IK, pose trajectories, SLERP trajectory interpolation, constrained/contact dynamics, controller classes, exact analytical workspace boundaries, or manipulability-colored workspace maps.