ROS Index
|
uav_simulator package from navrl repomap_manager navigation_runner onboard_detector uav_simulator |
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | MIT |
| Build type | CATKIN |
| Use | RECOMMENDED |
Repository Summary
| Description | [IEEE RA-L'25] NavRL: Learning Safe Flight in Dynamic Environments (NVIDIA Isaac/Python/ROS1/ROS2) |
| Checkout URI | https://github.com/zhefan-xu/navrl.git |
| VCS Type | git |
| VCS Version | main |
| Last Updated | 2025-04-23 |
| Dev Status | UNKNOWN |
| CI status | No Continuous Integration |
| Released | UNRELEASED |
| Tags | reinforcement-learning robotics collision-avoidance robot-navigation nvidia-isaac embodied-ai isaac-sim ros1-noetic ros2-humble |
| Contributing |
Help Wanted (0)
Good First Issues (0) Pull Requests to Review (0) |
Package Description
Additional Links
Maintainers
- Zhefan Xu
Authors
Lightweight Gazebo/ROS-based Simulator for Unmanned Aerial Vehicles (UAVs)
This package implements a lightweight quadcopter unmanned aerial vehicles (UAVs) simulator including various static and dynamic based on Gazebo/ROS. It also includes an optional PX4-based quadcopter simulation wrapper.
Author: Zhefan Xu, Computational Engineering & Robotics Lab (CERLAB) at Carnegie Mellon University (CMU).
If you find this work helpful, kindly show your support by giving us a free ⭐️. Your recognition is truly valued.
This repo can be used as a standalone package and also comes as a module of our autonomy framework.
I. Installation Guide
This repo has been tested on ROS Melodic with Ubuntu 18.04 and ROS Noetic with Ubuntu 20.04.
a. Non-PX4 Simulator (Required)
To install the non-PX4 simulator, please follow the standard catkin package make process as follows:
sudo apt-get install ros-[melodic/noetic]-mavros* # this package depends on mavros related ROS packages
git clone https://github.com/Zhefan-Xu/uav_simulator.git
cd ~/catkin_ws
catkin_make
setup environment variable. Add following to your
source path/to/uav_simulator/gazeboSetup.bash
b. PX4-based Simulator Wrapper (Optional but Recommended)
Please make sure that you have follow the previous steps to build the non-px4 simulator.
Step 1: Install vehicle models and make it compatible with your current ROS. The following lines give the summaries:
🍎 Current PX4 version has some problems with offboard mode, please use v1.12.0 as modified in the following lines:🍎
cd directory/to/install # this should not be your catkin workspace
git clone https://github.com/PX4/PX4-Autopilot.git --recursive --branch v1.12.0
bash ./PX4-Autopilot/Tools/setup/ubuntu.sh # this step will ask you to reboot
# Please make sure you reboot after the previous step
cd /path/to/PX4-Autopilot
DONT_RUN=1 make px4_sitl_default gazebo
Step 2: Add the following script to
```. Remember to replace
```<PX4-Autopilot_clone>
``` with the path to your PX4-Autopolot directory. This step will you setup the environment variables properly.
source
**Step 3**: Install geographiclib datasets for PX4 simulation.
wget https://raw.githubusercontent.com/mavlink/mavros/master/mavros/scripts/install_geographiclib_datasets.sh sudo bash ./install_geographiclib_datasets.sh
## II. Quick Start
a. To launch the non-PX4 simulator with a quadcopter:
roslaunch uav_simulator start.launch
You should be able to see a customized quadcopter in a predefined gazebo environment as shown below:
b. To launch the PX4 simulator with a quadcopter:
roslaunch uav_simulator px4_start.launch
You should be able to see a PX4 IRIS quadcopter in a predefined gazebo environment as shown below:
## III. Keyboard Control
Our non-PX4 customized simulator supports the keyboard control function. You are able to control the quadcopter motion **when you click the keyboard controller panel** shown as below:
## IV. Simulation Environments
- There are various predefined environments in this package and you can easily switch environment when you modify the launch file located in
```uav_simululator/launch/start.launch
``` or
```uav_simululator/launch/px4_start.launch
```. All the predifined environments are listed in the lanuch files.
- There are some environments which contains dynamic objects (e.g. moving persons). You can distinguish those dynamic environments by the environments' names. For example, the environment name
```floorplan1_dynamic_16.world
``` indicates that there are 16 dynamic objects in the floorplan1 environment.
One example of the dynamic environment is shown as below:
## V. ROS Topics
Here lists some important ROS topics related to the simulator:
- **Non-PX4 Simulator:**
-
```/CERLAB/quadcopter/cmd_acc
```: The command acceleration to the quadcopter.
-
```/CERLAB/quadcopter/pose
```: The ground truth pose of the quadcopter.
-
```/CERLAB/quadcopter/odom
```: The ground truth odom of the quadcopter.
-
```/CERLAB/quadcopter/setpoint_pose
```: The command pose to the quadcopter.
-
```/camera/color/image_raw
```: The color image from the onboard camera.
-
```/camera/depth/image_raw
```: The depth image from the onboard camera.
-
```/camera/depth/points
```: The depth cloud from the onboard camera.
- **PX4 Simulator**
-
```/mavros/setpoint_raw/attitude
```: The command to the quadcopter.
-
```/mavros/local_position/pose
```: The ground truth pose of the quadcopter.
-
```/mavros/local_position/odom
```: The ground truth odom of the quadcopter.
-
```/mavros/setpoint_position/local
```: The command pose to the quadcopter.
-
```/camera/color/image_raw
```: The color image from the onboard camera.
-
```/camera/depth/image_raw
```: The depth image from the onboard camera.
-
```/camera/depth/points
```: The depth cloud from the onboard camera.
## VI. Random World Generation
The simulator supports random (dynamic) world generation using the following commands:
Go to the generator directory
cd uav_simulator/scripts
Run random world generator
python3 generate_random_world.py
The parameter configuration file is in
```uav_simulator/scripts/world_generator.yaml
By default, the generated world will be saved under
uav_simulator/worlds/generated_env/generated_env.world
. Please add this to
start.launch
file to launch it for experiments.
The example generated environment is visualized:
https://github.com/user-attachments/assets/b6f2c4a0-5868-4f07-ab32-48e59bbc814f
The generated world also support map generation for map_manager and dynamic obstacle detection by onboard_detector. The map pcd files can be generated by turning on the
generate_map: True
parameter.
The example of the generated environment with the generated map and dynamic obstacles are shown below:
https://github.com/user-attachments/assets/22b5e256-2eb1-423c-b000-9d54690603b1
Wiki Tutorials
Package Dependencies
| Deps | Name |
|---|---|
| roscpp | |
| rospy | |
| std_msgs | |
| geometry_msgs | |
| sensor_msgs | |
| message_generation | |
| gazebo_ros | |
| mavros | |
| mavros_extras | |
| catkin | |
| message_runtime |
System Dependencies
Dependant Packages
Launch files
- launch/rviz.launch
- launch/px4_start.launch
-
- x [default: 0]
- y [default: 0]
- z [default: 0]
- R [default: 0]
- P [default: 0]
- Y [default: 0]
- est [default: ekf2]
- vehicle [default: iris]
- world [default: $(find uav_simulator)/worlds/generated_env/generated_env_static.world]
- sdf [default: $(find uav_simulator)/urdf/px4_iris.sdf]
- gui [default: true]
- debug [default: false]
- verbose [default: false]
- paused [default: false]
- respawn_gazebo [default: false]
- fcu_url [default: udp://:14540@localhost:14557]
- respawn_mavros [default: false]
- interactive [default: true]
- launch/start.launch
- launch/rviz_px4.launch
- launch/setupTF.launch
-
- pose_topic [default: /CERLAB/quadcopter/pose]