This repository contains Python code for the paper "Autonomous Navigation in Unknown Environments with Sparse Kernel-based Occupancy Mapping"

SCOPE

For simplicity, the code is for a 10x10 simulated environment instead of the ROS environment used in the paper. However, the main algorithms are the same.

DEPENDENCIES

The code depends on the following software and packages: python3.x, numpy, matplotlib, queue, rtree

COMMAND TO RUN THE CODE

python main.py

FILES

Important code

1. main.py: code for autonomous mapping and navigation algorithm
2. perceptron.py: code related to main contributions: Fastron and collision checking algorithms.
3. robot.py: code for collecting observations and retraining kernel perceptron

Others

4. env.py: code to generate a simulated environment
5. kernels.py: code for multiple kernel functions.
6. astar.py: code for A*
7. viz.py: code for visualization

VISUALIZATION EXPLAINED

1. Start cell: top left corner.
2. Goal cell: bottom right corner.
3. Green triangle: robot.
4. Blue boundary: decision boundary by Fastron score.
5. Magenta dashed boundary: inflated boundary by the upper bound.
6. Blue arrows: A* path.
7. Yellow arrows: traveled path.

TEST

The code has been tested on Ubuntu 16.04 and Python 3.6.