Path Planning and Control

In this project, we implement a number of path planning and control algorithms. Differential Drive and Bicycle motion models are developed for simulating the robot.

Usage

Note: This project has only been tested on Ubuntu 16.04 and Ubuntu 18.04

If setting up for the first time, follow these instructions to run the demos:

1. Install python dependencies:

   • If using conda (install + activate environment):

       conda env create -f environment.yml -n path-planning
       conda activate path-lanning
     

   • If using pip (requires python 3.6):

       pip install -r requirements.txt
     

2. Navigate to the working directory of the demos:

    cd path_planning
   

Path Planning and Control Demos

Potential Field

Guide a differential drive robot to a goal point using the potential field algorithm. A laser scanner is simulated to find distances to obstacles using Bresenham's rasterization algorithm.

Run the demo:

    python pf_demo.py

Probabilistic Roadmap (PRM)

Path planning using the Probabilistic Roadmap to 4 waypoints and control of a differential drive to each waypoint (and achieving a particular heading at each waypoint). A KDTree is implemented for an efficient generation of the roadmap.

Run the demo:

    python prm_demo.py

Rapidly-exploring Random Trees (RRT)

A bidirectional tree search version of RRTs (RRT-Connect) is implemented here, and a differential drive robot is controlled to visit 4 waypoints

Run the demo:

    python rrt_demo.py

Pure Pursuit Controller

Run the demo:

    python pure_pursuit_demo.py

Stanley Controller

Run the demo:

    python stanley_demo.py