Simple URDF Parser

A lightweight Python library for parsing, analyzing, and manipulating robot description files specified in the Unified Robot Description Format (URDF).

Features

• Load and parse URDF robot description files.
• Access link and joint names, origins (RPY + XYZ), and hierarchy.
• Forward kinematics (FK) computation.
• Inverse kinematics (IK) with Jacobian pseudoinverse and Damped Least Squares iterative solvers.
• Jacobian computation for configurations.
• Singular configuration detection.
• Generate random joint configurations respecting joint limits.
• Example script for quick demonstration of functionality.

Installation

Clone the repository and optionally install in editable mode:

git clone https://github.com/coenwerem/simple_urdf_parser.git
cd simple_urdf_parser
pip install -e .

Dependencies are listed in requirements.txt and include:

• numpy
• spatialmath-python
• trimesh

Usage

Parsing a URDF

This snippet assumes the URDF file is located in the assets/urdf folder of this repository. Otherwise, provide the absolute path to your URDF file.

from simple_urdf_parser.parser import Robot
import numpy as np
import spatialmath as sm

robot = Robot(desc_fp="../assets/urdf/ur3.urdf")

# List link and joint names
link_names = [link._name for link in robot.links]
joint_names = [joint.name for joint in robot.joints]
print("Links:", link_names)
print("Joints:", joint_names)

# Access joint origins (RPY + XYZ)
for joint in robot.joints:
    T = joint.origin.T
    print(joint.name, T.rpy(), T.t)

Forward Kinematics

q_test = robot.Configuration(
    joints=robot.actuated_joints,
    joint_values=[0, -np.pi/4, np.pi/4, 0, np.pi/3, 0]
)

T_fk = robot._compute_fk(
    config=q_test, 
    start=robot.base_link._name,
    end=robot.ee_link._name)
print("T_fk:", T_fk)

Inverse Kinematics

The current version supports IK computation using the Jacobian pseudo-inverse (method='jacinv') or Damped Least Squares (method='dls') methods. Use it like so:

ik_sol = robot._compute_ik(
    x_d=T_fk,
    init_guess=robot.Configuration.zeros_for_joints(robot.actuated_joints),
    method='jacinv',
    max_iters=200
)

# Verify
T_fk_sol = robot._compute_fk(
            config = robot.Configuration(
                    robot.actuated_joints,
                    joint_values=np.array(ik_sol.joint_values)
                    ))
print("IK solution:", np.round(np.array(ik_sol.joint_values), 3))
print("FK from IK solution:", T_fk_sol)

Jacobian and Singularity Checks

jac = robot._compute_jacobian(config=q_test)
print("Jacobian:\n", np.round(jac, 4))
print("Is singular:", q_test.is_singular(robot))

Random Configuration Sampling

q_rand = robot.Configuration.random_config(robot.actuated_joints, robot)
print("Random configuration:", np.round(qrand.joint_values, 4))

Examples

Run the included demo:

python3 examples/demo_parse_urdf.py

Sample output:

Links:
['base_link', 'base_link_inertia', 'shoulder_link', 'upper_arm_link', 'forearm_link',
 'wrist_1_link', 'wrist_2_link', 'wrist_3_link', 'base', 'flange', 'tool0']

Joint origins (RPY + XYZ) printed per joint...

T_fk:
  -0.5   -0.0  0.866  0.4565
   0.866  0.0  0.5    0.1533
   0.0    1.0 0.0    0.2388
   0      0   0      1

IK sol: [-0.0, -0.786, 0.786, -0.0, 1.047, -0.0]
T_fk (from IK): ...
Jacobian at q_test: ...
Random configuration: [-1.5766, 5.6638, 1.4577, 1.2398, -4.3226, -4.3229]

Project Structure

simple_urdf_parser/
├── assets/urdf/        # Sample URDF files
├── examples/           # Example scripts demonstrating usage
├── simple_urdf_parser/ # Main library module
│   └── parser.py
├── tests/              # Unit tests
├── pyproject.toml
├── requirements.txt
└── README.md

Notes

• The current implementation uses proxy boxes for mesh geometry due to deprecation of PyCollada. Avoid using it for collision-sensitive computations that rely on meshes.
• Designed for Python >= 3.9.
• Intended as a lightweight utility for robotics projects, simulation, and education.

License

MIT License. See LICENSE for details.

Issue Tracker

Bugs, feature requests, and other issues can be reported on the GitHub Issues page.

Please provide:

• A clear description of the problem or feature request.
• Steps to reproduce the issue.
• Relevant code snippets or URDF files.
• Python version and OS.

Detailed reports will be much appreciated to help improve the library!