FastSLAM

A particle-based FastSLAM algorithm in Python for pioneer 3dx mobile robot

This algorithm takes in a .2d format laser scan log and returns a 2d array grid map

Size of the map is bounded by the maximum and minimum scan coord in the scan hisotory. Thus, this is not an online slam algorithm.

Usage

1. obtain a .2d file from the robot using sickLogger
2. open slam.py, modify parameters like filename, num_particles, result_folder, etc.
3. run python3 slam.py

Files

slam.py: the main code for running the slam algorithm

particle_class.py: code for the particle class, including position update, map update, weight update

scan_match.py: scan-match algorithms, including icp and a customized algorithm called search_best_angle

utility.py: important utilities, including coord transfomation, map drawing, and ray tracing

jerodlab.2d: a sample .2d laser log of jerod's lab

baseline_map.png: a baseline map of jerod's lab constructed using only the laser log and without any slam correction

map_slam.png: a map of jerod's lab constructed using slam

map_naive.png: a map of jerod's lab constructed using slam but w/o scan-matching

progress_full: a folder containing the construction history of the slam map

Progress Demo

Baseline:

SLAM:

SLAM without Scan-Matching:

Warning

Current code assume the angle increment is 1. Some code might need changes if angle increment is 0.5