Article Dans Une Revue IEEE Access Année : 2025

A Modular Execution Architecture for Robust Multi-Robot Planning and Acting in Trans-Media Environments

Résumé

In complex missions involving heterogeneous multi-robot teams, especially in transmedia systems that operate across environments such as air and water, robust execution frameworks must ensure both temporal coherence and resilience to uncertainty. These challenges stem from the need to manage dynamic mode transitions, closely linked inter-agent tasks, and execution-time failures. This paper introduces the Adaptive and Modular Architecture (AMA) Execution and Planning components. AMA-EXEC is a distributed execution framework designed to enable coherent, fault-tolerant mission execution in such conditions. AMA-EXEC uses the plans produced by AMA-PLAN, a PDDL-based planning framework. AMA-EXEC incorporates Simple Temporal Networks (STNs) to facilitate temporal reasoning; modular BTs to enable distributed execution; and runtime monitoring mechanisms to categorize failures, propagate delays, and execute partial replanning. In contrast to centralized, monolithic systems, AMA-EXEC organizes execution around collaborative robot teams and leverages real-time feedback to maintain synchronization and temporal alignment under disturbances. The framework's validation process involves the execution of simulated trans-media missions, encompassing concurrent tasks and a range of failure scenarios. The findings indicate enhanced execution continuity, extended complex cases application, and robust failure recovery in comparison to baseline methodologies. The AMA-EXEC system's modularity and generalizability render it suitable for a wide range of applications, including environmental monitoring, distributed exploration, and search-and-rescue operations.

Fichier principal
Vignette du fichier
access_reviewed.pdf (730.7 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05244656 , version 1 (08-09-2025)
hal-05244656 , version 2 (01-10-2025)

Licence

Identifiants

Citer

Virgile de la Rochefoucauld, Photchara Ratsamee, Simon Lacroix, Félix Ingrand, Yuki Uranishi. A Modular Execution Architecture for Robust Multi-Robot Planning and Acting in Trans-Media Environments. IEEE Access, In press, ⟨10.1109/ACCESS.2025.3625646⟩. ⟨hal-05244656v2⟩
592 Consultations
224 Téléchargements

Altmetric

Partager

  • More