Awesome

motion-planners

Flexible python implementations of several robotic motion planners.

Citation

Caelan Reed Garrett. Motion Planners. https://github.com/caelan/motion-planners. 2017.

Example

A simple motion planning problem in 2D workspace and 2D configuration space that only depends on Tkinter.

motion-planners$ python -m motion_planners.tkinter.run

Algorithms

Single-Query

Sampling-Based:

• Rapidly-Exploring Random Tree (RRT)
• Bidirectional RRT (BiRRT/RRT-Connect)
• MultiRRT
• RRT*

<img src="images/rrt.png" height="200"> <img src="images/birrt.png" height="200"> <img src="images/rrt*.png" height="200">

Grid Search

• Breadth-First Search (BFS)

• A*
• Lattice

Other

• Straight-Line Path
• Linear Shortcutting

<img src="images/lattice.png" height="200"> <img src="images/rrt+shortcut.png" height="200">

Multi-Query

Sampling-based:

• Probabilistic Roadmap (PRM)
• Lazy PRM

<img src="images/degree_prm.png" height="200"> <img src="images/eager_prm.png" height="200"> <img src="images/lazy_prm.png" height="200">

API

Each motion planner takes as input python functions that perform its primitive operations, allowing them to be flexibly used in many software environments.

• :param distance_fn: Distance function - distance_fn(q1, q2)->float
• :param sample_fn: Sample function - sample_fn()->conf
• :param extend_fn: Extension function - extend_fn(q1, q2)->[q', ..., q"]
• :param collision_fn: Collision function - collision_fn(q)->bool

Applications

• PyBullet Motion Planning - https://github.com/caelan/pybullet-planning
• PyBullet Task and Motion Planning (TAMP) - https://github.com/caelan/pddlstream

<img src="images/turtlebot_motion.png" height="200"> <img src="images/pr2_motion.png" height="200"> <img src="images/se3.png" height="200">