Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Scalable Solution Methods for Dec-POMDPs with Deterministic Dynamics
Yang You, Alex Schutz, Zhikun Li, Bruno Lacerda, Robert Skilton, Nick Hawes
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| initialize with Det-POMDP heuristic (local observation space) instead of joint-observation centralized heuristics, then run IDPP iterations (ablation) | IDPP · MACTP <5,2,14> and Collecting <5,5,2,4> | discounted return | Improved |
| use Det-MCVI Det-POMDP best-response solver instead of Monte Carlo POMCP (MCJESP) | JESP framework · MACTP <4,2,12> | discounted return: 459.64 ± 104.58 -> 798.47 ± 2.40 | Improved |
| use IDPP instead of MCJESP | JESP framework · Collecting <5,5,2,4> | discounted return: 237.09 ± 32.80 -> 315.56 ± 2.81 | Improved |
| use IDPP instead of MCJESP | JESP framework · MACTP <5,2,14> | computation time (s): 4115.8 ± 659.6 -> 1706.6 ± 298.4 | Improved |
| use IDPP instead of exact InfJESP (SARSOP) on small instances | JESP framework · Collecting <4,3,2,2> | discounted return: 186.39 ± 0.72 -> 184.42 ± 1.16 | Hurt |
| use IDPP instead of InfJESP on large instances | JESP framework · MACTP <4,2,8> and larger | feasibility: InfJESP infeasible (time/memory limits) | Improved |
| use IDPP instead of MAA* | Dec-POMDP solver · MACTP <3,2,5> | memory usage | Improved |
| use IDPP instead of IQL | planner vs MARL · MACTP <4,2,12> | discounted return: 526.31 ± 64.60 -> 798.47 ± 2.40 | Improved |
| use IQL instead of MAPPO | MARL baseline · MACTP and Collecting instances | discounted return: 542.82 ± 46.48 -> 696.33 ± 45.92 | Improved |
| add iterative best-response refinement on top of Det-POMDP heuristic init (ablation) | Det-POMDP heuristic · MACTP <4,2,12> | discounted return: 603.95 ± 8.05 -> 798.47 ± 2.40 | Improved |
| add iterative best-response refinement on top of Det-POMDP heuristic init (ablation) | Det-POMDP heuristic · Collecting <4,3,2,2> | discounted return: 118.60 ± 2.11 -> 184.42 ± 1.16 | Improved |
Closest papers by meaning
- Optimally Solving Simultaneous-Move Dec-POMDPs: The Sequential Central Planning Approach (AAAI 2025 · similarity 0.88)
- Locally Interdependent Multi-Agent MDP: Theoretical Framework for Decentralized Agents with Dynamic Dependencies (ICML 2024 · similarity 0.87)
- Factored Online Planning in Many-Agent POMDPs (AAAI 2024 · similarity 0.87)
- Deterministic Policy Gradient Primal-Dual Methods for Continuous-Space Constrained MDPs (AAAI 2025 · similarity 0.86)
- Learning Logic Specifications for Policy Guidance in POMDPs: an Inductive Logic Programming Approach (AAAI 2025 · similarity 0.86)
- Multi-Environment POMDPs with Finite-Horizon Objectives (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
