airsim_img_publisher

Overview

This repository is a fork of the publishAirsimImgs repository by Marcelino Almeida. It is a preliminary solution to integrate ROS and AirSim plugin till Microsoft provides an alternative one.

Features within this repo:

Publishes rgb data into the topic /airsim/rgb/image_raw
Publishes depth data into the topic /airsim/depth
Publishes surface normals data into the topic /airsim/normals/image_raw
Publishes segmentation labels data into the topic /airsim/segmentation/image_raw
Publishes camera calibration parameters into /airsim/depth/camera_info
Publishes a tf tree with the origin (world), the position/orientation of the quadcoper (base_link), and the position/orientation of the camera (camera_frame)

The airsim_img_publisher package has been tested under ROS Kinetic and Ubuntu 16.04LTS. The source code is released under MIT Licence.

Changelog

urdf model of DJI M100

Added urdf model of a quadrotor similar to the quarotor’s blueprint in the simulator
The header files have been renamed and shifted to the include directory
Fixed bugs in the code
README modified for better understanding
Another transform has been added from the drone’s base_frame_id frame to camera_frame_id
Changed the names of topics that are published
Added publishing of normals and segmentation images as well
Script added to fly the drone in lawn-mower surveillance pattern
Added separate node to publish stereo camera images only

Installation

Building from Source

Dependencies

Eigen
```
sudo apt-get install libeigen3-dev
```
AirSim: In order to link the client-side of the plugin with this project, build the fork of the plugin available here
```
cd ~/
git clone https://github.com/Mayankm96/AirSim_ROS.git
./setup.sh
./build.sh
```

Building

To build from source, clone the latest version from this repository into your catkin workspace
```
cd ~/catkin_ws/src
https://github.com/Mayankm96/airsim_img_publisher.git
```
In order to run AirSim, you will have to change the Airsim_ip and Airsim_port parameters to match the IP/Ports in which Airsim is running. All these informations can be found in the settings.json file (located at ~/Documents/AirSim) for your Airsim configuration. The ports you are looking for are the “LogViewerPort” and the “UdpPort”. Note that the settings.json file have to be configured such that “LogViewerHostIp” and “UdpIp” both have the IP of the computer that will run AirSim.
Set the correct path to AIRSIM_ROOT in the CMakeLists.txt file.
To compile the package:
```
cd ~/catkin_ws
catkin_make
```

Usage

Before running the nodes in the package, you need to run Airsim plugin in the Unreal Engine. In case you are unfamiliar on how to do so, refer to the tutorials available here.

Running the `tf` publisher of drone model (DJI M100)

If you wish to use the urdf model of the drone used in AirSim simulator, then

Change tf_cam_flag to false in pubImages.launch and pubStereoImages.launch

Run:

roslaunch airsim_img_publisher publish_tf.launch

Alternately, you may set tf_cam_flag to true and allow the node to publish the required transformations on it’s own

NOTE: In the modified blueprint of the drone for UE4, all cameras are downward-facing.

Running image publisher

Change the IP and Port configurations in pubImages.launch to match the settings in which Airsim is running. Then:

roslaunch airsim_img_publisher pubPointCloud.launch

Create octomap

roslaunch airsim_img_publisher octomap.launch

Visualization on rviz

An rviz configuration file can be found in the rviz directory. This configuration allows a user to see the published images, as well as the tf tree.

rosrun rviz rviz -d ~/catkin_ws/src/airsim_img_publisher/rviz/pclConfig.rviz

Nodes