AI & ML
MA-SafeDiffuser: Safe Multi-Agent Planning with Diffusion Probabilistic Models
arXiv cs.MA0 views
arXiv:2611.05524v1 Announce Type: cross Abstract: We propose MA-SafeDiffuser, a multi-agent extension of SafeDiffuser that equips diffusion-based trajectory planners with finite-time diffusion invariance guarantees for joint-agent safety specifications. Building on the single-agent construction that embeds control barrier function (CBF) constraints into reverse-diffusion updates, we: (i) formalize joint safe sets as intersections of per-agent and pairwise barrier sets; (ii) derive centralized and decentralized (communication-aware) constrained denoising procedures with provable invariance under mild assumptions; (iii) address local-trap and deadlock phenomena via time-varying specifications and liveness CBFs; and (iv) develop a lightweight benchmarking suite including a multi-agent Maze2D domain. Empirically, MA-SafeDiffuser reduces violation counts relative to unconstrained diffusion baselines, while retaining planning quality. We provide algorithms, proofs, and reference implementations.
Source：https://arxiv.org/abs/2611.05524PreviousWho May Overrule the Agent? Evidence-Gated Authority in LLM Review InstitutionsNextDecision Models as Scalable Economic Surrogates that Capture Human Variation
Related posts
Scalable Source Selection for Machine Learning Tasks
0arXiv cs.DB
AvalancheBench: Evaluating Enterprise Agent-Driven Semantic Query Processing Through Latent World Recovery
0arXiv cs.DB
ConStruM: A Structure-Guided LLM Framework for Context-Aware Schema Matching
0arXiv cs.DB
From Certified Pairs to a Certified Hub: Reasoner-Checked Crosswalks for the Information Exchange Standard
0arXiv cs.DB
A Comparative Study of Importance Attribution to Columns in Deep Learning over Relational Databases
0arXiv cs.DB
Community Search under Member-Specific Tolerances
0arXiv cs.DB