Running SUMMIT

Launching simulator

Open a terminal in the folder where SUMMIT was extracted, and run ./CarleUE4.sh to start the simulator. A window will open, containing an empty map.

The simulator is now running as a server, waiting for client apps to connect and to interact (e.g. spawning map objects dynamically, simulating a crowd, etc.) with it.

Running an example scenario

Here, we provide a walkthrough of how to use SUMMIT to run a simple scenario, to showcase the various features of SUMMIT.

You will spawn a map of Meskel Square, one of SUMMIT's built-in maps. Next, you will spawn a heterogeneous crowd on the map. Finally, you will run a sample agent, which attempts to navigate through the crowd.

Firstly, spawn the Meskel Square map using the built-in spawn_meshes.py and spawn_imagery.py scripts. This will spawn the meshes and imagery relevant to Meskel Square:

<summit_root>/PythonAPI/examples/spawn_meshes.py --dataset meskel_square
<summit_root>/PythonAPI/examples/spawn_imagery.py --dataset meskel_square

Secondly, spawn a heterogeneous crowd on the map using the built-in gamma_crowd.py script. This will simulate a heterogeneous traffic on the spawned map:

<summit_root>/PythonAPI/examples/gamma_crowd.py --dataset meskel_square

Finally, run the built-in sample meskel_square_ego_vehicle.py script. This will spawn an agent that navigates through the crowd.

<summit_root>/PythonAPI/examples/meskel_square_ego_vehicle.py

What's next?

To get started writing your own scripts to interact with SUMMIT, you may step through the programming tutorials:

• Load and spawn existing maps
• Using roads and sidewalks

To learn how to prepare your own maps, or to read more about how to use the built-in traffic simulator, you may step through the map tutorials:

• Prepare new maps
• Simulate traffic in new maps
• Configuring the simulation

To use the expert planner, check out our Context-POMDP repository.