A Matlab motion planner ensemble of a global Voronoi model and a local Potential Field model
Matlab C++
Switch branches/tags
Nothing to show
Permalink
Failed to load latest commit information.
.gitignore
MyHeap.h
README.markdown
aStar.m
anneal.m
calccost.m
drawCircle.m
grad2.m
graddesc.m
localplan.m
makeNoiseFilter.m
noiseFilter.m
noiseFilterDist.m
obstacle2.mat
obstacles.mat
plotlocal.m
plotlocal3d.m
pq_create.cpp
pq_create.m
pq_create.mexglx
pq_create.mexw64
pq_delete.cpp
pq_delete.m
pq_delete.mexglx
pq_delete.mexw64
pq_demo.m
pq_pop.cpp
pq_pop.m
pq_pop.mexglx
pq_pop.mexw64
pq_push.cpp
pq_push.m
pq_push.mexglx
pq_push.mexw64
pq_size.cpp
pq_size.m
pq_size.mexglx
pq_size.mexw64
pq_top.cpp
pq_top.m
pq_top.mexglx
pq_top.mexw64
runFigs.m
setupN1000.mat
simulator_main.m
testcost.m
voronoi_astar.m
voronoi_planner.m

README.markdown

Simultaneous Planning Localization And Mapping For Unmanned Aerial Vehicles

Authors: Bill, Kyel, Sam, Will. Final Class Report

Published in ICRA: Kyel Ok, Sameer Ansari, William Gallagher, William Sica, Frank Dellaert, Mike Stilman, "Path Planning with Uncertainty: Voronoi Uncertainty Fields." IEEE International Conference on Robotics and Automation 2013.

This project is a MATLAB Simulator for UAV navigation with vision uncertainty using SLAM. The goal is to implement this on a raptor quadcopter Spring 2012, with the intent to use in search and rescue operations in forested environments.

More Images & Videos

Final simulation Traveled Path

The build progress is documented on the class wiki.


The goal of the project is to implement a planning technique into the SLAM framework, using a local planner to direct the path towards the global exploration goal. This work is motivated to improve SLAM performance in uncertain or rapidly changing environments, such as search and rescue missions in a forest using a UAV.

This simulator is written as part of a research assignment for the GeorgiaTech course Robotics Intelligence & Planning (Fall 2011).

The simulator is written as a script with user defined variables, accessible through simulator_main.m

Global Planner

A Global planner is implemented using a Voronoi diagram with DFS or A* search through nodes to the global goal. This is updated each turn based on uncertainty and UAV/robot position

Local Planner

A local planner implemented using potential field with gradient descent. Only visible (within a range of UAV) obstacles are considered. This is updated each turn based on uncertainty and UAV/robot position