pith. sign in

arxiv: 2601.20969 · v3 · submitted 2026-01-28 · 💻 cs.AI

The Epistemic Planning Domain Definition Language: Official Guideline

Alessandro Burigana , Francesco Fabiano This is my paper

Pith reviewed 2026-05-16 10:25 UTC · model grok-4.3

classification 💻 cs.AI
keywords epistemic planningEPDDLDynamic Epistemic LogicPDDLabstract event modelsmulti-agent planningknowledge and belief
0
0 comments X

The pith

EPDDL supplies a PDDL-style language that encodes the full semantics of Dynamic Epistemic Logic for epistemic planning tasks.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces EPDDL to replace the fragmented ad hoc representations currently used for epistemic planning problems. It grounds the language in Dynamic Epistemic Logic through a new device called abstract event models, which describe how actions change what agents know and believe. A reader would care because the single formal language lets different planners accept the same input, lets researchers share benchmarks without translation, and removes one barrier to systematic progress in multi-agent planning that treats knowledge as a first-class concern.

Core claim

EPDDL provides a unique PDDL-like representation that captures the entire DEL semantics, enabling uniform specification of epistemic planning tasks. The authors develop abstract event models as the semantic foundation for actions and give a complete formal syntax and semantics for the language that stays faithful to DEL. Examples drawn from representative benchmarks demonstrate how the language supports interoperability and reproducible evaluation without the ad hoc extensions common in prior work.

What carries the argument

Abstract event models, a compact representation for epistemic actions that specifies exactly how each action updates agents' knowledge and beliefs while preserving the full expressive power of DEL.

If this is right

  • Planners built for different DEL fragments can now accept identical problem files for direct performance comparison.
  • Benchmark suites can be written once and reused across research groups without translation layers.
  • New epistemic planners can be developed against a single, documented standard rather than against private input formats.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A common language makes it feasible to build hybrid systems that combine epistemic planners with classical PDDL solvers on subproblems.
  • Standard benchmarks could accelerate the development of planners that handle richer DEL fragments without each group starting from scratch.
  • The same abstract event model mechanism might later be extended to continuous or probabilistic belief updates while keeping backward compatibility.

Load-bearing premise

Abstract event models can represent any epistemic action so that the full expressive power of DEL is retained without loss or the need for custom extensions.

What would settle it

An epistemic planning scenario whose action effects on knowledge and belief cannot be expressed inside the abstract event model format without omitting information that DEL would distinguish.

Figures

Figures reproduced from arXiv: 2601.20969 by Alessandro Burigana, Francesco Fabiano.

Figure 1
Figure 1. Figure 1: Initial state 𝐼 of Example 1. Each block is represented by a rounded square and is labeled by its name. Column names are placed below the gray line. the form 𝑐 ◁ 𝑒, where 𝑐 ∈ ℒ𝑃 𝑟𝑜𝑝 is a formula and 𝑒 is a set of literals of P. We say that 𝑎 is applicable in a classical state 𝑆 iff 𝑆 |= Pre(𝑎), and, if so, the update of 𝑎 in 𝑆 is the classical state 𝑆 ∘ 𝑎 = (𝑆 ∖ Eff− 𝑆 (𝑎)) ∪ Eff+ 𝑆 (𝑎), where Eff𝑆 (𝑎) = {… view at source ↗
Figure 2
Figure 2. Figure 2: Epistemic state of Example 2. Rectangles represent worlds, labelled by their name (the actual world has a double line). Edges denote the accessibility relation. We represent labels graphically, similarly to Example 1, e.g., 𝐿(𝑤1) = {𝑜𝑛(𝑏1, 𝑐1), 𝑜𝑛(𝑏2, 𝑏1), 𝑜𝑛(𝑏3, 𝑐2), 𝑜𝑛(𝑏4, 𝑐3)}. A propositional atom is true in a world if it is contained in the label of that world. The clauses of propositional connectives… view at source ↗
Figure 3
Figure 3. Figure 3: Epistemic state of Example 3. Since we now have multiple agents, each edge is labeled by the corresponding agent(s). ℒP,Ag : (𝑀, 𝑤) |= 𝑝 iff 𝑝 ∈ 𝐿(𝑤) (𝑀, 𝑤) |= ¬𝜑 iff (𝑀, 𝑤) ̸|= 𝜑 (𝑀, 𝑤) |= 𝜑 ∧ 𝜓 iff (𝑀, 𝑤) |= 𝜑 and (𝑀, 𝑤) |= 𝜓 (𝑀, 𝑤) |= □𝑖𝜑 iff for all 𝑣 ∈ 𝑊, if 𝑤𝑅𝑖𝑣 then (𝑀, 𝑣) |= 𝜑 Example 3 (Multi-agent Epistemic Blocks World). Consider the situation of Example 2 and let 𝑎 be the agent that can only se… view at source ↗
Figure 4
Figure 4. Figure 4: Non-deterministic epistemic state of Example 5. On top of the internal perspectives of agents, multi-pointed epistemic models can also be employed to represent non-determinism, as we now show. The disjoint union of two states 𝑠 = ((𝑊, 𝑅, 𝐿), 𝑊𝑑) and 𝑠 ′ = ((𝑊′ , 𝑅′ , 𝐿′ ), 𝑊′ 𝑑 ) is the epistemic state 𝑠⊔𝑠 ′ = ((𝑊 ⊔𝑊′ , 𝑅⊔ 𝑅′ , 𝐿⊔𝐿 ′ ), 𝑊𝑑 ⊔𝑊′ 𝑑 ), where ⊔ denotes the disjoint union of two sets. Non-determ… view at source ↗
Figure 5
Figure 5. Figure 5: Public announcement 𝑎 = 𝑎𝑛𝑛(□𝑟¬𝑜𝑛(𝑏1, 𝑐3)) (left) and epistemic state 𝑠 ′ 𝑟 = 𝑠𝑟 ⊗ 𝑎 (right) of Example 7. Events are graphically represented by squares (designated events are boxed). Each event 𝑒 is labeled by a pre-/postconditions pair ⟨pre(𝑒), post(𝑒)⟩ (if 𝑒 has trivial postconditions, we simply label it with pre(𝑒)). 𝑒 nil 𝑙 𝑎, 𝑙, 𝑟 𝑎, 𝑟 𝑏1 𝑏3 𝑏2 𝑏4 𝑐1 𝑐2 𝑐3 (𝑣1, e) 𝑏1 𝑏2 𝑏3 𝑏4 𝑐1 𝑐2 𝑐3 (𝑣1, nil) 𝑏2 𝑏1… view at source ↗
Figure 6
Figure 6. Figure 6: Private ontic action privMove(𝑙, 𝑏2, 𝑏1, 𝑏3) and state 𝑠 ′′ 𝑟 = 𝑠 ′ 𝑟 ⊗ privMove(𝑙, 𝑏2, 𝑏1, 𝑏3). The preconditions and postconditions of the events are described in Example 8. are (𝑤1, 𝑒) and (𝑤2, 𝑒). Since 𝑒 has trivial postconditions, applying 𝑒 does not change the labels of worlds. Finally, since 𝑄𝑖 = {(𝑒, 𝑒)} for all 𝑖 ∈ Ag, the update preserves all edges between the new worlds. We then have 𝑠𝑟 ⊗ 𝑎 |= … view at source ↗
Figure 7
Figure 7. Figure 7: Quasi-private sensing action quasiPrivPeek(𝑎, 𝑟, 𝑏2, 𝑏1) and epistemic state 𝑠 ′′ 𝑟 = 𝑠 ′ 𝑟 ⊗ quasiPrivPeek(𝑎, 𝑟, 𝑏2, 𝑏1). The pre- and postconditions of the events are described in Example 8. • 𝐸 = {𝑒, 𝑓, nil}, 𝐸𝑑 = {𝑒, 𝑓}; • 𝑄𝑖 = {(𝑒, 𝑒),(𝑓, 𝑓),(nil, nil)}; • 𝑄𝑗 = {(𝑒, 𝑒),(𝑒, 𝑓),(𝑓, 𝑒),(𝑓, 𝑓),(nil, nil)}; • 𝑄𝑘 = {(𝑒, nil),(𝑓, nil),(nil, nil)}; • pre(𝑒) = Clear (𝑏) ∧ On(𝑏, 𝑥); • pre(𝑓) = Clear (𝑏) ∧ ¬On(𝑏… view at source ↗
Figure 8
Figure 8. Figure 8: Epistemic state 𝑠 (left), non-deterministic action 𝑎 (middle) and state (𝑠 ′ = 𝑠⊗𝑎) of Example 10. Event 𝑒 represents the move of 𝑏1 on top of 𝑏4 and 𝑓 the move of 𝑏4 on top of 𝑏1. Event 𝑒 represents the case where the coin lands on heads, triggering the move of 𝑏1 to 𝑏4, and event 𝑓 represents the opposite instance. Preconditions and postconditions for moving blocks are the standard ones. As the result of… view at source ↗
Figure 9
Figure 9. Figure 9: Pointed abstract frame for private actions, where 𝑓 and 𝑜 are observability types denoting fully observant and oblivious agents, respectively. Definition 15 (Abstract Frames). Let ObsTypes be a finite, non-empty set of observability types. An abstract frame on ObsTypes is a pair 𝐹 = (𝐸, 𝑄), where: • 𝐸 ̸= ∅ is a finite set of events; and • 𝑄 : ObsTypes → 2 𝐸×𝐸 assigns to each observability type 𝑡 ∈ ObsTypes… view at source ↗
Figure 10
Figure 10. Figure 10: State 𝑠 ′ 𝑟 ⊗ absPrivMove(𝑙, 𝑏2, 𝑏1, 𝑏3), where 𝑠 ′ 𝑟 is the epistemic state from Example 7 and the action is from Example 12. Example 13 (Abstract Product Update). Let 𝑠 ′ 𝑟 = ((𝑊′ , 𝑅′ , 𝐿′ ), 𝑊′ 𝑑 ) be the epistemic state from Example 7. It is a local state for agents 𝑎 and 𝑟 representing the fact it is commonly known to all agents that both 𝑎 and 𝑟 don’t know whether block 𝑏1 is under 𝑏2 or 𝑏3. We ass… view at source ↗
Figure 11
Figure 11. Figure 11: Induced epistemic action absPrivMove(𝑙, 𝑏2, 𝑏1, 𝑏3)↓𝑠 ′ 𝑟 from Example 14. set of abstract actions of ℒ 𝐶 P,Ag , and 𝜑𝑔 ∈ ℒ𝐶 P,Ag is the goal formula. A solution (or plan) to 𝑇 is a finite sequence 𝜋 of abstract actions of Act such that: 1. 𝜋 is applicable in 𝑠0; and 2. 𝑠0 ⊙ 𝜋 |= 𝜑𝑔. 3.1. Expressivity Abstract epistemic actions can be systematically converted into standard actions (Definition 10), and vic… view at source ↗
read the original abstract

Epistemic planning extends (multi-agent) automated planning by making agents' knowledge and beliefs first-class aspects of the planning formalism. One of the most well-known frameworks for epistemic planning is Dynamic Epistemic Logic (DEL), which offers an rich and natural semantics for modelling problems in this setting. The high expressive power provided by DEL make DEL-based epistemic planning a challenging problem to tackle both theoretically, and in practical implementations. As a result, existing epistemic planners often target different DEL fragments, and typically rely on ad hoc languages to represent benchmarks, and sometimes no language at all. This fragmentation hampers comparison, reuse, and systematic benchmark development. We address these issues by introducing the Epistemic Planning Domain Definition Language (EPDDL). EPDDL provides a unique PDDL-like representation that captures the entire DEL semantics, enabling uniform specification of epistemic planning tasks. Our main contributions are: 1. A formal development of abstract event models, a novel representation for epistemic actions used to define the semantics of our language; 2. A formal specification of EPDDL's syntax and semantics grounded in DEL with abstract event models. Through examples of representative benchmarks, we illustrate how EPDDL facilitates interoperability, reproducible evaluation, and future advances in epistemic planning.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The paper introduces the Epistemic Planning Domain Definition Language (EPDDL), a PDDL-like formalism for specifying epistemic planning tasks. It defines abstract event models as a novel representation for epistemic actions and supplies a formal syntax and semantics grounded in Dynamic Epistemic Logic (DEL), with the goal of enabling uniform task specification, interoperability across planners, and reproducible benchmark evaluation.

Significance. If the central claim holds, EPDDL would address a genuine fragmentation problem in epistemic planning by providing a standardized, DEL-complete representation language. This could support systematic comparisons, shared benchmarks, and progress on implementation challenges that currently arise from ad-hoc encodings of different DEL fragments.

major comments (1)
  1. [formal development of abstract event models (contribution 1) and semantics specification (contribution 2)] The claim that abstract event models capture the entire DEL semantics (including arbitrary higher-order epistemic updates) is load-bearing for the main contribution, yet the manuscript provides no explicit embedding theorem, bijection, or preservation proof relating abstract event models to standard DEL event models. Without such a result, it remains open whether every DEL construct can be expressed without loss or ad-hoc extensions.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their detailed review and valuable comments on our manuscript introducing EPDDL. We address the major comment regarding the formal development of abstract event models below.

read point-by-point responses

  1. Referee: [formal development of abstract event models (contribution 1) and semantics specification (contribution 2)] The claim that abstract event models capture the entire DEL semantics (including arbitrary higher-order epistemic updates) is load-bearing for the main contribution, yet the manuscript provides no explicit embedding theorem, bijection, or preservation proof relating abstract event models to standard DEL event models. Without such a result, it remains open whether every DEL construct can be expressed without loss or ad-hoc extensions.

    Authors: We agree that an explicit embedding result would clarify the relationship. Abstract event models are defined to directly mirror the structure of DEL event models, with preconditions, effects, and accessibility relations specified in a way that allows a straightforward translation to standard DEL semantics. The semantics of EPDDL are given by mapping to these models and applying the standard product update from DEL. This ensures by construction that the full DEL semantics, including higher-order epistemic updates, are captured without loss. To address the referee's concern, we will include an explicit theorem stating the equivalence (bijection) between abstract event models and standard DEL event models, along with a preservation proof, in the revised manuscript. revision: yes

Circularity Check

0 steps flagged

EPDDL is a definitional language specification with no circular derivation

full rationale

The paper introduces EPDDL as a new PDDL-like language by explicitly defining abstract event models as a novel representation and then providing a formal syntax and semantics grounded in DEL. No load-bearing steps reduce by construction to inputs, fitted parameters, or self-citations; the work consists of formal development and specification rather than predictions or uniqueness theorems derived from prior author results. The central claim of capturing DEL semantics uniformly is achieved through direct definitional grounding without any self-referential reduction or renaming of known results.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the domain assumption that DEL semantics can be captured via a PDDL-like syntax and the newly introduced abstract event models, with no free parameters or data-fitted elements.

axioms (1)
  • domain assumption Dynamic Epistemic Logic offers rich and natural semantics for modeling epistemic planning problems.
    The language is explicitly grounded in DEL as stated in the abstract.
invented entities (1)
  • abstract event models no independent evidence
    purpose: Novel representation for epistemic actions used to define the semantics of EPDDL.
    Introduced in the paper as part of the formal development to enable the language specification.

pith-pipeline@v0.9.0 · 5513 in / 1120 out tokens · 28663 ms · 2026-05-16T10:25:39.529401+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

  1. [1]

    Ghallab, D

    M. Ghallab, D. S. Nau, P. Traverso, Automated planning - theory and practice, Elsevier, 2004

    work page 2004

  2. [2]

    Russell, P

    S. Russell, P. Norvig, Artificial Intelligence: A Modern Approach (4th Edition), Pearson,

    work page

  3. [3]

    URL: http://aima.cs.berkeley.edu/

    work page

  4. [4]

    Hintikka, Knowledge and belief: An introduction to the logic of the two notions, Studia Logica 16 (1962) 119–122

    J. Hintikka, Knowledge and belief: An introduction to the logic of the two notions, Studia Logica 16 (1962) 119–122

    work page 1962

  5. [5]

    Bolander, M

    T. Bolander, M. B. Andersen, Epistemic Planning for Single and Multi-Agent Systems, J. Appl. Non Class. Logics 21 (2011) 9–34. URL: https://doi.org/10.3166/jancl.21.9-34. doi: 10. 3166/jancl.21.9-34

    work page doi:10.3166/jancl.21.9-34 2011

  6. [6]

    Blackburn, M

    P. Blackburn, M. d. Rijke, Y. Venema, Modal Logic, Cambridge Tracts in Theoretical Computer Science, Cambridge University Press, 2001. doi:10.1017/CBO9781107050884

    work page doi:10.1017/cbo9781107050884 2001

  7. [7]

    D. K. Lewis, Convention: A Philosophical Study, Wiley-Blackwell, Cambridge, MA, USA, 1969

    work page 1969

  8. [8]

    R. J. Aumann, Agreeing to disagree, The Annals of Statistics 4 (1976) 1236–1239

    work page 1976

  9. [9]

    Lismont, P

    L. Lismont, P. Mongin, On the logic of common belief and common knowledge, Theory and Decision 37 (1994) 75–106. doi:10.1007/bf01079206

    work page doi:10.1007/bf01079206 1994

  10. [10]

    J. Y. Halpern, Y. Moses, Knowledge and common knowledge in a distributed environment, J. ACM 37 (1990) 549–587. URL: https://doi.org/10.1145/79147.79161. doi:10.1145/79147. 79161

    work page doi:10.1145/79147.79161 1990

  11. [11]

    Baltag, L

    A. Baltag, L. S. Moss, S. Solecki, The Logic of Public Announcements and Common Knowledge and Private Suspicions, in: I. Gilboa (Ed.), Proceedings of the 7th Conference on Theoretical Aspects of Rationality and Knowledge (TARK-98), Evanston, IL, USA, July 22-24, 1998, Morgan Kaufmann, 1998, pp. 43–56

    work page 1998

  12. [12]

    van Eijck, Guarded Actions, 2004

    J. van Eijck, Guarded Actions, 2004. Tech. rep., Centrum voor Wiskunde en Informatica, Amsterdam, CWI Report SEN-E0425

    work page 2004

  13. [13]

    van Benthem, J

    J. van Benthem, J. van Eijck, B. P. Kooi, Logics of Communication and Change, Inf. Comput. 204 (2006) 1620–1662. doi:10.1016/j.ic.2006.04.006

    work page doi:10.1016/j.ic.2006.04.006 2006

  14. [14]

    H. P. van Ditmarsch, W. van der Hoek, B. P. Kooi, Dynamic Epistemic Logic, volume 337, Springer Netherlands, 2007

    work page 2007

  15. [15]

    Aucher, T

    G. Aucher, T. Bolander, Undecidability in epistemic planning, in: F. Rossi (Ed.), IJCAI 2013, Proceedings of the 23rd International Joint Conference on Artificial Intelligence, Beijing, China, August 3-9, 2013, IJCAI/AAAI, 2013, pp. 27–33. URL: http://www.aaai.org/ ocs/index.php/IJCAI/IJCAI13/paper/view/6903

    work page 2013

  16. [16]

    D. E. Smith, D. S. Weld, Conformant graphplan, in: J. Mostow, C. Rich (Eds.), Pro- ceedings of the Fifteenth National Conference on Artificial Intelligence and Tenth Inno- vative Applications of Artificial Intelligence Conference, AAAI 98, IAAI 98, July 26-30, 1998, Madison, Wisconsin, USA, AAAI Press / The MIT Press, 1998, pp. 889–896. URL: http://www.aa...

    work page 1998

  17. [17]

    Baral, G

    C. Baral, G. Gelfond, E. Pontelli, T. C. Son, An Action Language for Multi-Agent Domains: Foundations, CoRR abs/1511.01960 (2015). URL: http://arxiv.org/abs/1511.01960

    work page arXiv 2015

  18. [18]

    Ghallab, A

    M. Ghallab, A. Howe, C. Knoblock, D. Mcdermott, A. Ram, M. Veloso, D. Weld, D. Wilkins, PDDL—The Planning Domain Definition Language, 1998. URL: http://citeseerx.ist.psu.edu/ viewdoc/summary?doi=10.1.1.37.212

    work page 1998

  19. [19]

    Plaza, Logics of public announcements, in: Proceedings 4th International Symposium on Methodologies for Intelligent Systems, 1989

    J. Plaza, Logics of public announcements, in: Proceedings 4th International Symposium on Methodologies for Intelligent Systems, 1989

    work page 1989

  20. [20]

    Plaza, Logics of public communications, Synth

    J. Plaza, Logics of public communications, Synth. 158 (2007) 165–179. URL: https://doi. org/10.1007/s11229-007-9168-7. doi:10.1007/S11229-007-9168-7

    work page doi:10.1007/s11229-007-9168-7 2007

  21. [21]

    Kominis, H

    F. Kominis, H. Geffner, Beliefs in multiagent planning: From one agent to many, in: R. I. Brafman, C. Domshlak, P. Haslum, S. Zilberstein (Eds.), Proceedings of the Twenty-Fifth International Conference on Automated Planning and Scheduling, ICAPS 2015, Jerusalem, Israel, June 7-11, 2015, AAAI Press, 2015, pp. 147–155. URL: http://www.aaai.org/ocs/index. p...

    work page 2015

  22. [22]

    C. J. Muise, V. Belle, P. Felli, S. A. McIlraith, T. Miller, A. R. Pearce, L. Sonenberg, Plan- ning over multi-agent epistemic states: A classical planning approach, in: B. Bonet, S. Koenig (Eds.), Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelli- gence, January 25-30, 2015, Austin, Texas, USA, AAAI Press, 2015, pp. 3327–3334. URL: ht...

    work page doi:10.1609/aaai.v29i1.9665 2015

  23. [23]

    Huang, B

    X. Huang, B. Fang, H. Wan, Y. Liu, A general multi-agent epistemic planner based on higher- order belief change, in: C. Sierra (Ed.), Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI 2017, Melbourne, Australia, August 19-25, 2017, ijcai.org, 2017, pp. 1093–1101. URL: https://doi.org/10.24963/ijcai.2017/152. ...

    work page doi:10.24963/ijcai.2017/152 2017

  24. [24]

    Muise, V

    C. Muise, V. Belle, P. Felli, S. A. McIlraith, T. Miller, A. R. Pearce, L. Sonenberg, Efficient multi-agent epistemic planning: Teaching planners about nested belief, Artif. Intell. 302 (2022) 103605. URL: https://doi.org/10.1016/j.artint.2021.103605. doi:10.1016/J.ARTINT. 2021.103605

    work page doi:10.1016/j.artint.2021.103605 2022

  25. [25]

    T. C. Son, E. Pontelli, C. Baral, G. Gelfond, Finitary s5-theories, in: E. Fermé, J. Leite (Eds.), Logics in Artificial Intelligence - 14th European Conference, JELIA 2014, Funchal, Madeira, Portugal, September 24-26, 2014. Proceedings, volume 8761 ofLecture Notes in Computer Science, Springer, 2014, pp. 239–252. URL: https://doi.org/10.1007/978-3-319-115...

    work page doi:10.1007/978-3-319-11558-0_17 2014