Arena-rosnav-3D is structurally very similar to arena-rosnav to allow easy switching between 2D and 3D environments. Currently, three custom worlds and a random world generator are supported. In our detailed custom worlds you can generate dynamic obstacles either in a scenario mode or in a random mode where the obstacle trajectories are determined randomly.
We provide the following worlds for your use:
 |
 |
 |
 |
 |
 |
|---|---|---|---|---|---|
| aws_house | turtlebot3_house | small_warehouse | random world | factory | hospital |
 |
 |
 |
|---|---|---|
| experiment_rooms | bookstore | turtlebot3_world |
Furthermore, an outside world is available using the flag "outside:=true", e.g.
roslaunch arena_bringup start_arena_gazebo.launch outside:=true
| Worlds | adjusted_mean | mean | variance | Entropy | Max-Entropy | MapSize | MinObsDis | NumObs | OccupancyRatio | obs_ratio |
|---|---|---|---|---|---|---|---|---|---|---|
| eval_floot | 0.054 | 317.647 | 5918.087 | 0.366 | 2.322 | 4000 | 0.224 | 18 | 0.0498 | 0.45 |
| factory | - | 360 | 0 | 0.104 | 2.322 | 36000 | 0.223 | 70 | 0.009 | 0.194 |
| hospital | 1.328 | 358.998 | 270.379 | 0.175 | 2.322 | 64000 | 0.224 | 82 | 0.019 | 0.128 |
| ignc | 0.008 | 180 | 21406.596 | 0.146 | 2.322 | 4840 | 1000 | 1 | 0.017 | 0.020 |
| small_warehouse | 0.008 | 200.930 | 26624.237 | 0.265 | 2.322 | 6250 | 0.224 | 30 | 0.032 | 0.48 |
| aws_house | 0.182 | 351.628 | 1926.969 | 0.308 | 2.322 | 4000 | 0.224 | 41 | 0.037 | 1.025 |
| bookstore | 0.006 | 114.545 | 18997.649 | 0.193 | 2.322 | 9000 | 0.224 | 59 | 0.019 | 0.655 |
| exp_room | 0.033 | 270 | 8106.014 | 0.095 | 2.322 | 9000 | 1000 | 1 | 0.009 | 0.011 |
| outside | - | 360 | 0 | 0.112 | 2.322 | 10240 | 1000 | 1 | 0.012 | 0.009 |
| tb3_house | - | 360 | 0 | 0.089 | 2.322 | 12250 | 0.4 | 5 | 0.008 | 0.040 |
| tb3_world | 0.006 | 180 | 29083.099 | 0.068 | 2.322 | 7372.8 | 0.75 | 10 | 0.006 | 0.135 |
the explanation of each metric can be found at hier
To select one of our pre-build worlds, specify your scenario by selecting aws_house, turtlebot3*house, small_warehouse *(bold is your default world)_ in your startup file:
roslaunch arena_bringup start_arena_gazebo.launch world:=turtlebot3_houseSince our pre-build worlds are very detailed we do not spawn further static obstacles. For obstacle and robot task-management see here.
To implement your own custom world file see here
To select our custom world generator specify the outside argument with true (default is false)
roslaunch arena_bringup start_arena_gazebo.launch outside:=trueThe custom world generator creates a map with a set number of static and dynamic obstacles of a randomly determined size (and shape).
If you want to change the number of static obstacles you can also do that via the terminal (the shape is randomly determined by the task-generator node):
roslaunch arena_bringup start_arena_gazebo.launch outside:=true num_static_obs:=10If you want to change the number of actors you can specify the amount using parameters like this:
roslaunch arena_bringup start_arena_gazebo.launch outside:=true agents:=4By combining the random 2d map generation feature from our own arena-tools with the seamless image to Gazebo world conversion of LIRS_World_Construction_Tools we can test our navigation approaches on magnitude of 3D worlds, varying in layout, complexity. For more information on how to use this feature please refer to arena-tools. Otherwise, if you already have your own map image in mind, visit LIRS_World_Construction_Tools to gain information on how to convert it into a Gazebo world.
We support different robots:
 |
 |
 |
 |
|---|---|---|---|
| turtlebot3_burger | jackal | ridgeback | agv-ota |
 |
 |
 |
 |
|---|---|---|---|
| Robotino(rto) | turtlebot3_waffle_pi | Car-O-Bot4 (cob4) | dingo |
All robots are equipped with a laser scanner. The robots differ in size, laser-range etc. See below table for more detailed information on each robot:
| Name | Max Speed (vx) [_m/s] | Max Speed (vy) [_m/s] | Max Rotational Speed (θy) [_rad/s] | Radius [m] | Emergency-Stop¹ | Laser-range [m] | Holonomic² |
|---|---|---|---|---|---|---|---|
| turtlebot3-burger | 0.22 | 0.0 | 2.84 | 0.113 | True | 3.5 | False |
| jackal | 2.0 | 0.0 | 4.0 | 0.267 | True | 30.0 | False |
| ridgeback | 1.1 | 0.5 | 2.0 | 0.625 | True | 10.0 | True |
| agv-ota | 0.5 | 0.0 | 0.4 | 0.629 | True | 5.0 | False |
| rto³ | 2.78 | todo | todo | 0.225 | todo | todo | todo |
| turtlebot3_waffle_pi | 0.26 | 0.0 | 1.82 | 0.208 | False | 3.5 | False |
| Car-O-Bot4 (cob4)³ | 1.1 | todo | todo | 0.36 | True | todo | todo |
| dingo | 1.3 | 0.0 | (4.0) | 0.378 | todo | 30.0 | False |
For additional / more detailed information about each robot:
- See the parameters needed for the Navigation stack
- See additional robot parameters like laser min/max [rad]
- See HERE for the definition of the robots action_spaces (needed for rl-based-training)
NOTE: The emergency-stop capability is currently still being development, however, it will be available on all robots.
To select a robot model for your simulation run (in this case ridgeback):
roslaunch arena_bringup start_arena_gazebo.launch model:=ridgeback¹ Stops moving when an object has been detected in the safety zone of the robot
² For non-holonomic robots vel_y = 0; they are not able to drive directly to their left or right, but have to drive on a circular trajectory to their reach navigation goal
³ We only use the navigation platform of these robots
Arena-rosnav-3d supports different local planner and task-modes. Select your planner to move to the respective documentation.
Note, that not every robot can be used with every planner, yet. See an overview on possible cominations in the table below:
| Model | TEB | DWA | MPC | ROSNAV | CADRL | RLCA | Gring | Navrep | ARENA | All-in-One |
|---|---|---|---|---|---|---|---|---|---|---|
| burger | X | X | X | X | X | X | X | X | X | X |
| jackal | X | X | X | X | X | X | X | X | X | |
| ridgeback | X | X | X | X | X | X | ||||
| agv-ota | X | X | X | X | X | |||||
| rto | X | X | X | X | X | X | X | X | ||
| youbot | X | X | X | (X) | X | X | X | X | ||
| waffle_pi | X | X | X | X | X | |||||
| cob4 | X | X | X | X | X | X | ||||
| dingo | X | X | X | X |
Take note of the different task modes, which define how the start and goal position of the robot, as well as the trajectories of the obstacles will be set. Here we support the task modes:
-
random: Here the start and goal position of the robot and the trajectories of the dynamic obstacles is determined randomly by arena-rosnavs task-generator node.
roslaunch arena_bringup start_arena_gazebo.launch task_mode:=random
-
scenario: Here start and goal position of the robot, as well as the trajectory of dynamic obstacles are predefined in scenario files, which need to be selected at start up. You can define you own scenarios using the arena-tools repository. Here to start the scenario mode:
roslaunch arena_bringup start_arena_gazebo.launch task_mode:=scenario scenario_file:=small_warehouse_obs05.json
-
manual: Here you can set the start and goal position of the robot manually, using rviz
Goal Posebutton. Dynamic obstacles are not spawned.roslaunch arena_bringup start_arena_gazebo.launch task_mode:=manual
You can start these planners by setting the local_planner parameter in the launch file. As example, here how to start teb:
roslaunch arena_bringup start_arena_gazebo.launch local_planner:=tebNOTE: The planer rlca is currently only available for the robots: turtlebot3_burger, turtlebot3_waffle_pi
When first using the cadrl, refer to the documentation here since you will need to create a different virtual environment.
To start the cadrl planner run:
workon cadrl
roslaunch arena_bringup start_arena_gazebo.launch local_planner:=cadrlWhen first using rosnav make sure to run the following commands:
pip install gym
cd ~/catkin_ws/src/forks/stable-baselines3 && pip install -e .When first using navrep and guldenring make sure to follow the install documentation here first.
| Name | Type | Default | Description |
|---|---|---|---|
| task_mode | random | scenario | random | The task mode denotes if the robot start and goal position, as well as the trajectory of the dynamic obstacles is set randomly In scenario mode all values are set beforehand. We currently support one scenario per world. To extend this and build your own custom scenarios see here. |
| local_planner | teb | dwa | mpc | rlca | cadrl | arena | rosnav | rosnav | Denotes which planner is used to set the roboters cmd_vel |
| Name | Type | Default | Description |
|---|---|---|---|
| network_type | rosnav | navrep | guldenring | rosnav | This is mainly needed to pick the right encoder. You can find a detailed documentation of how to add a new encoder and what this parameter exactly means here |
| trainings_environment | rosnav | navrep | guldenring | rosnav | Indicates in which trainings environment the model you want to test has been trained. For all inputs except rosnav the drl node is not started automatically. |
| agent_name | string | tb3 | Must exactly match the name of your model file. The model file needs to be in a directory named like the trainings_environment. More on this here |
| model | turtlebot3_burger | jackal | ridgeback | agv-ota | turtlebot3_burger | All roboter models we provide. Each roboter model has different velocities and scan sizes. Therefore, it is crucial that the model parameter matches the roboter your model is for. |
If you want to have a rosnav model called rosnav_model control the turtlebot3_burger robot your program call should look like this:
roslaunch arena_bringup start_arena_gazebo.launch model:=turtlebot3_burger local_planner:=rosnav world:=map5 scenario_file:=map5_obs_05.json agent_name:=turtlebot3_burger trainings_environment:=rosnav network_type:=rosnavWhereas the arguments trainings_environment and network_type can be omitted for this use case, cause they are set by default.
You can further customize the simulation altering the following parameters:
| Name | Options | Description |
|---|---|---|
| actors | Integer | Specify the amount of actors to be inserted into a Gazebo world. All of the actors position are then managed by the pedsim simulator |
| actor_height | Double | A fixed actor height on a per world basis needs to be set, default value of 1.1 is already set and should be appropriate for most of the worlds, if not experimentation for proper value will be needed |
| additional_map | true, false | Set to true, if you wish to use an additional, more detailed occupancy map for obstacles/pedsim manager |
| add_map_path | String | Specify the path to the yaml file of the additional map |
| train_mode | true, false | Since training mode is not yet implemented false should stay selected. (This would deactivate e.g. the task_generator) |
| disable_scenario | true, false | This parameter would e.g. disable the task generator and other selected notes. This should only be changed with caution |
| reset | true, false | Determines whether the scenario should be reset (or terminated) after reaching its objective (Robot has reached navigation goal) |
| enable_pedsim | true, false | Pedsim is used to for dynamic obstacle management. Setting this to false should disable dynamic obstacles |
| enable_pedvis | true, false | Responsible for converting pedsim messages, such as agent states, obstacles and providing their visualization in Rviz |
<arg name="train_mode" default="false"/>
<arg name="local_planner" default="dwa" doc = "local_planer type [tep, dwa, mpc, rlca, cadrl, arena, rosnav]"/>
<arg name="rviz_file" default="nav_LP"/>
<arg name="disable_scenario" default="false"/>
<arg name="reset" default="true"/>
<arg name="enable_pedsim" default="true"/>
<arg name="outside" default="false"/>
<arg name="world" default="aws_house" unless="$(arg outside)"/>
<arg name="world" default="outside" if="$(arg outside)"/>
<arg name="model" default="turtlebot3_burger"
<arg name="scenario_file" default="$(arg world).json"/>
<arg name="step_size" default="0.01"/>
<arg name="update_rate" default="250"/>
<arg name="show_viz" default="true"/>
<arg name="viz_pub_rate" default="30" />
<arg name="use_rviz" default="true"/>
<arg name="map_path" default="$(find simulator_setup)/maps/$(arg world)/map.yaml" />
<arg name="task_mode" default="random"/>
<arg name="obs_vel" default="0.3" />
<arg name="action_frequency" default="10"/>
<arg name="global_frame_id" value="map"/>
<arg name="odom_frame_id" value="odom"/>
<arg name="base_frame_id" value="base_footprint"/>
<arg name="odom_ground_truth" default="/odometry/ground_truth"/>There are a few things to consider when including additional navigation apparatus:
- The laser scan msg is published under the
/scantopic. Note if the laser did not register any obstacles, the gazebo simulator fills the scan msg withinfvalues, wereas the Flatland simulator fills it withnanvalues - The robot position is published under the
/odomtopic, the velocity under the/cmd_vel - In a simple 2D environment like Flatland, many costmap-plugins will not be usable. Make sure to check, whether a local and global costmap is being published