NRSL Autonomous UAV Project

Description

This repository contains the code for UAV to implement autonomous navigation in an unknow enviorment.

Features

Using visual-inertial odometer to estimate uav pose.
Build a octree map in real time using stereo camera, and more importantly, this process dose not require GPU support.
Implemented a frotier-based exploration strategy and RRT*-based trajectory planning algorithm.
The established octree map and the trajectory of the uav can be optimized offline.
Once the exploration is complete, the uav can be relocalized to improve location accuracy and use the optimized octree map for path planning

videos

Installation

Install ROS Kinetic

Ubuntu 16.04. ROS Kinetic. ROS Installation

Install maplab

You must install maplab and maplab_dependence in branch "pre_release_public/july-2018", not master branch

Install maplab plugin: octomap-maplab-plugin

    cd ~/maplab_ws/src
    git clone https://github.com/songjin321/octomap-maplab-plugin.git
    catkin build --no-deps octomap_maplab_plugin

Install Our Autonomous Navigation WorkSpace

Clone the repository and catkin build:

    cd ~
    git clone https://github.com/HKUST-Aerial-Robotics/VINS-Mono.git
    cd ./uav_navigation
    catkin build
    source ~/uav_navigation/devel/setup.bash

How to Run it

Prerequisite

sensors
- a stereo camera such as zed camera, publish raw image in topic /cam0/image_raw and /cam1/image_raw
- a imu sensor such as xsen imu, publish raw imu information in topic /imu0
move_base, provide a service, let uav fly to goal pose
validate device, motion capture system, publish true pose in topic /true_pose

exploration

# terminal 1 
source ~/Project/uav_navigation/devel/setup.bash
roslaunch application exploration.launch

# terminal 1 
source ~/Project/uav_navigation/devel/setup.bash
roslaunch application mavros.launch

# terminal 2 rovio in vio mode
source ~/Project/maplab_ws/devel/setup.bash
./Project/maplab_ws/src/maplab/applications/rovioli/scripts/tutorials/huang_live ~/Documents/maps

# terminal 3 record data
rosbag record -j -b 0 /cam0/image_raw /cam1/image_raw /imu0 /vrpn_pose -O ~/Documents/exploration.bag

rosrun octomap_server octomap_saver ~/Documents/explorationMap.bt

# terminal 4 rviz show
rviz ~/Project/uav_navigation/src/application/config/expolration.rviz

off-line optimization

# replay expolration bag (optional)
rosbag play ~/Documents/exploration.bag --clock
roslaunch application run_bag.launch
./Project/maplab_ws/src/maplab/applications/rovioli/scripts/tutorials/huang_live

# import point cloud resource
source ~/Project/maplab_ws/devel/setup.bash
rosrun resource_importer import_resources_w_ncamera_yaml.sh ~/Documents/maps ~/Documents/pointclouds.bag /point_cloud ~/Documents/zed.yaml ~/Documents/maps

# optimize the map and save refined point cloud to file
rosrun maplab_console maplab_console
load --map_folder ~/Documents/maps_pc
v
rtl
optvi
lc
optvi
save --map_folder ~/Documents/maps_opt
**rosparam set /use_sim_time false**
create_octomap

# use octomap_server creat new map
rosbag play ~/Documents/pointcloud_processed.bag

source ~/Project/uav_navigation/devel/setup.bash
roslaunch application run_bag.launch

rosrun octomap_server octomap_saver ~/Documents/optimizedMap.bt

relocalization demo

# terminal 1 
source ~/Project/uav_navigation/devel/setup.bash
roslaunch application localization.launch

# terminal 2 rovio in localization mode
source ~/Project/maplab_ws/devel/setup.bash
./Project/maplab_ws/src/maplab/applications/rovioli/scripts/tutorials/huang_localization ~/Documents/maps_localization ~/Documents/maps_loc

# terminal 3 record data
rosbag record -j -b 0 /cam0/image_raw /cam1/image_raw /imu0 /vrpn_pose -O ~/Documents/localization.bag

# terminal 4 rviz show
rviz ~/Project/uav_navigation/src/application/config/localization.rviz

xiangyudxy / uav_navigation