-
Notifications
You must be signed in to change notification settings - Fork 121
Demo Scenario
To demonstrate the use of the planner, demo scenarios in simulation are included in the provided package. The space considered for exploration is a 14x14m flat, where one piece is separated by a wall with a narrow passage.
Firefly hexacopter in the RotorS simulator, performing exploration of a flat.
To run the demo scenario, make sure to follow the installation instructions and then run
roslaunch interface_nbvp_rotors flat_exploration.launch
This launches the planner, as well as the gazebo based simulator RotorS, simulating an instance of a firefly hexacopter. You can display the exploration progress in rviz following the visualization guid. The result should look similar to this:
Initially the robot is performing some motion to get an initial map, where enough free space is available to plan the first step.
After some steps, the known area has grown already.
Most of the area has already been explored, except for the separated part, which is only accessible through a small gap in the wall.
The planner finds this gap reliably and navigates towards it.
After exploration of the separated room, the planner aims to fill in the small missing regions.
As the unexplored parts of the scenario become sparse, the tree has to grow denser to find a non-zero gain.
A second demo scenario is provided, using the same environment. This time, the surface of the walls, floor and ceiling has to be explored. For this purpose, a mesh was derived from the first demo scenario's dense reconstruction and loaded to the planner. The gain at a certain configuration is then computed by the visible area of not previously seen parts of the mesh. To run the demo, type
roslaunch interface_nbvp_rotors area_exploration.launch
To run the two scenarios as a multi agent mission, type:
roslaunch interface_nbvp_rotors multiagent_flat_exploration.launch
or
roslaunch interface_nbvp_rotors multiagent_area_exploration.launch
To visualize the planning process, refer to the Visualization wiki page.
Contact details: Andreas Bircher, Kostas Alexis, Autonomous Systems Lab