Downloads

Currently two versions of GRT and one version of MO-GRT are available.

Version 1.0

This is the version that has been used in the AIPS-00 Competition.

Click here to get it.

Click here to get the domains of the AIPS-00 competition.

Click here to get the domains of the AIPS-98 competition.

Note that GRT participated at the un-typed STRIPS track.

Version 2.0

This is the latest version (however it is still under development). The version supports state-constraints and resources (in case you use both of them, the version may not work properly). Moreover, it supports some new features (limited object elimination, new methods to enrich the goal state etc) that improve the search process.  

Click here to get the version 2.0.

Click here to get the AIPS-00 planning competition problems.

Click here to get the AIPS-98 planning competition problems. 

Click here to get some logistics problems with XOR-constraints (from the AIPS-98 competition).

Click here to get some simplified mystery problems with XOR-constraints.

Click here to get the 'mystery' problems of the AIPS-98, translated in a more comprehensive language and transformed so that they support resources.

MO-GRT

This is the new multi-objective version of GRT, which has the ability to plan for multiple criteria simultaneously.

Here is the source code of MO-GRT (version: December 2001).

Here is the executable program (for Sun Solaris).

Here are some problems from the logisticsMO domain, a logistics domain with many resources.

Here are some script files to run MO-GRT in the logisticsMO problems.

You can get some preliminary results (in MS-Excel 2000 format or unformatted).

Instructions

We are mainly working in MS-Windows platform and Microsoft Visual C++, version 6.0.

If you are using the same environment, just create a project in C++, add the files to the project and build them.

If you are working in unix, unzip the zipped file in a directory and just write make.

There are several command line options. These are:

bullet-d domain_file :  Defines the domain file.
bullet-p problem_file : Defines the  problem file.
bullet-o output file : Defines an output file (optional).
bullet-strategy best : Defines that a best-first search strategy will be used (optional).
bullet-strategy hill : Defines that a hybrid hill-climbing and breadth-first search strategy, similar to the FF search strategy, will be used (optional). Note that if no strategy is defined, the best-first will be used.
bullet-os win: Defines that the current operating system is MS-Windows (optional).
bullet-os win: Defines that the current operating system is unix (optional). Note that is OS declaration is omitted, it is assumed that the OS is MS-Windows.
bullet-echo1: Displays limited messages in the screen, while running.
bullet-echo: Displays too-many messages in the screen, while running. These are mainly debugging messages.
bullet-r0: Switch that determines the way the enriched goal state is constructed. It favors facts that can be combined with goal facts, in order to make actions applicable to the goals.
bullet-r2: Switch that determines the way the enriched goal state is constructed. It favors facts that are members of the initial state. Note that if no one of the -r0 and -r2 switches is used, then all the candidate facts, i.e. facts that are not mutexed with any goal fact, are considered goal facts.
bullet-time N: Specifes the maximum time limit (in minutes), during which the problem has to be solved. If this time is passed, the execution stops.

Example: Run GRT planner with the following command:

grt -d "domain.pddl" -p "probLOGISTICS-10-0.pddl" -os win -time 10 -r0 -echo1

 

Please e-mail us for any problem or suggestion, in case you have downloaded the GRT planner. In case you want to be informed about new versions of the GRT planner, send us an e-mail.

 

Last updated: 25/02/2004 , by Ioannis Refanidis