This repository contains my planning algorithm research code. It has modules for creating random or dynamic occupancy grids on which to generate plans, utilities for plotting and animating plans, and implementations of the RRT, RRT*, and RRT-Informed planners.
Utility Modules:
- Random World Generation with Perlin Noise
- Animation Module
- Dubins Primitive Module
Planning Modules:
- RRT Planner
- RRT(star) Planner
- RRT(informed) Planner
- Dubins Vehicle RRT Planner
- Dubins Vehicle RRT(star) Planner
Path Following Modules:
- Straight Line Vector-Field Path-Follower
- Decomposition of multi-waypoint plan for vector-field follower using voronoi regions
All algorithms are written in Python. I've tried to accelerate some things with Numba; so, your platform needs to be compatible with numba. But for the most part, it's standard python, and as such, is extremely slow.
This is research code, so it's not at all a robust package to use for a real planner. Let me know how you are using it!
It relies on Numba to accelerate some collision logic:
https://numba.readthedocs.io/en/stable/user/installing.html
pyfastnoisesimd
relies on SIMD instructions available only on modern processors to speed up perlin noise generation for random occupancyGrids. I'm not sure how it degrades if those instructions are not available. See https://pyfastnoisesimd.readthedocs.io/en/latest/overview.html.
RRT:
Planners:
Path Followers: