This domain is described in [Crosby, Jonsson and Rovatsos, 2014]. In this case, there are two different domains depending on the multiagent notation used:
maze_dom_cal.pddl
for [Kovacs, 2012] notation.maze_dom_cn.pddl
for [Crosby, Jonsson and Rovatsos, 2014] notation.
The problems in the problems
folder can be related to any of the domains. Besides, inside this folder there is a folder called generator
containing C++ code for creating new instances. You can compile it as follows:
g++ generate.cpp -o generate
The usage of the generator is the following:
generate <agents> <iter> <lo> <hi> <step> <door> <bridge> <boat> <switch>
where:
agents
is the number of agents.iter
is the number of instances for each combination of<number of agents, grid size>
.lo
andhi
are the minimum and maximum size of the grid respectively. The generator will create an instance forlo
size. Then it will progressively increase this quantity bystep
,2 * step
,3 * step
, ... until reaching or surpassinghi
.door
: percentage of opened doors in the maze.bridge
: percentage of bridges in the maze.boat
: percentage of boats in the maze.switch
: percentage of switches (with an associated locked door) in the maze.
-
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.
-
Crosby, M., Jonsson, A., and Rovatsos, M. (2014). A Single-Agent Approach to Multiagent Planning. Proceedings of the 21st European Conference on Artificial Intelligence (ECAI-14), 237-242.