DLVK

This is the homepage of DLVK, which implements a frontend for the language K to the DLV system.

Features

The language K is a logic-based planning language which - unlike previously proposed languages of this kind - is well-suited for representing incomplete knowledge. For a thorough description of the language we refer to "Planning under Incomplete Knowledge" by Thomas Eiter, Wolfgang Faber, Nicola Leone, Gerald Pfeifer, and Axel Polleres.

Graphical User Interface

For better readability of plans and the output of our planning frontend Jürgen Oberhofer has created a GUI available at SourceForge written in Tcl/TK.

Publications and Further Information

Publications on DLVK and some more information on our planning project can be found at the official project page.

Examples

We suggest to run some of them to get familiar with the system.

• Blocksworld domain
  Sussman anomaly instance (background knowledge)
  command line: DLV blocks_world.plan sussman.plan sussman.dl -FP -N=3
• Blocksworld domain assuring correct initial states
  Instance with incomplete initial state (background knowledge)
  command line: DLV blocks_world_assured.plan sussman_incomplete.plan sussman_incomplete.dl -FP -N=4
• Yale Shooting
  command line: DLV yale.plan -FP -N=1
• Monkey and Banana (background knowledge)
  command line: DLV monkey.plan monkey.dl -FP -N=4
• Monkey and Banana with incomplete initial state (background knowledge)
  command line: DLV monkey_incomplete.plan monkey.dl -FP -N=4
• Rocket (background knowledge)
  command line: DLV rocket.plan rocket.dl -FP -N=3

Some more examples (these encodings were used for benchmarks in Technical Report INFSYS RR-1843-01-12 "A Logic Programming Approach to Knowledge-State Planning, II: The DLV^K System"):

• Blocksworld: domain description, background knowledge, problem descriptions (P1,P2, P3,P4, P5)
  command line: e.g. DLV blocks.plan P.bk P3.plan -N=7 -FP -planlength=8
  Use option -N to specify the number of blocks, e.g. -N=7 means that there are 8 blocks numbered 0,...,7 in this example
• Bomb in toilet : domain descriptions (BT(p), BTC(p), BTUC(p), BMTC(p,t), BMTUC(p,t), concurrency.plan, noConcurrency.plan), background knowledge (bt1.bk(one toilet), bt2.bk(two toilets))
  e.g. command line for BMTC(4,2) with concurrent actions:
  DLV bmtc.plan concurrency.plan bt2.bk -N=4 -FPsec -planlength=3
  Use option -N to specify the number of packages, e.g. -N=4 means that there are 4 packages numbered 1,...,4 in this example
• Bomb in toilet (knowledge-based encodings): domain descriptions (BTkb(p), BTCkb(p), BTUCkb(p), BMTCkb(p,t), BMTUCkb(p,t))

Even more examples (some showing the new cost optimal planning features of our system which will be presented at JELIA'02):

• Travelling Salesperson with exceptional costs (background knowledge)
  command line: DLV tsp.plan tsp.bk -FP -N=10
• Crossing the Bridge (background knowledge)
  command line: DLV crossing.plan crossing.bk -FP -planlength=7
• Sheep and Wolves (background knowledge)
  command line: DLV sheep_wolves.plan sheep_wolves.bk -FP
• Sokoban
  An encoding of sokoban in K, with some example levels packed in a gzipped tar file. Refer to the README file for details.