This domain is described in [Boutilier and Brafman, 2001]. The domain folder contains a domain generator called generate_domain.py. The difference between two TableMover domains is given by the number of tables in the problem since tables are used as constants. The generator creates domains using [Kovacs, 2012] notation, and it is used as follows:

python generate_domain.py <num-tables> <add-move-agent-action>

where num-tables is the number of tables in the domain, and add-move-agent-action indicates whether to add or not the action for moving agents between rooms (i.e. no need to be move the table to go from one room to another).

In the same domain folder you can find TableMover domains ranging from 1 to 4 tables (table_domain1.pddl, ..., table_domain4.pddl).

The problems folder also contains a generator inside the generator subfolder. The script is called generate_instance.py and it is used as follows:

python generate_instance.py <num-nodes> <num-edges> <num-blocks> <num-agents> <num-tables> <instance-number>

where:

• num_nodes is the number of rooms.
• num_edges is the number of links between rooms (they are randomly created).
• num_agents is the number of agents.
• num_tables is the number of tables.
• instance_number is a number used to differentiate between two instances with the same number of nodes, edges, ...

• Boutilier, C. and Brafman, R. I. (2001). Partial-Order Planning with Concurrent Interacting Actions. Journal of Artificial Intelligence Research (JAIR) 14, 105-136.

• Kovacs, D. L. (2012). A Multi-Agent Extension of PDDL 3.1. In Proceedings of the 3rd Workshop on the International Planning Competition (IPC), 19–27.