A Closed-Loop Framework-Independent Bridge from AIPlan4EU's Unified Planning Platform to Embedded Systems
Résumé
The main goal of the AIPlan4EU project is to give easy and unified access to a large and diversified number of planning technologies and approaches. Yet, to make beneficial use of the produced plans, the project also develops and deploys Technology Specific Bridges (TSB) to connect and close the loop with real-world applications. In this paper we present one of these bridges which specifically targets robotic applications and more generally embedded systems. In these systems, planning and plan execution lead to commands on effectors in the real world, while sensors, intrinsic and extrinsic, report on the state of the world, which is then fed back to monitor the plan execution, with replanning or plan repair when needed. To this effect, we have developed a Python library that simplifies the creation of planning domains for existing applications and allows to execute and monitor resulting plans. Despite being developed for robotic applications, the library is not dependent on a specific robot framework, nor any middleware, and can also be used for non-robotic domains. We demonstrate this TSB on two robotic experiments: a ROS (Quigley et al. 2009) based service robot and a GenoM (Dal Zilio et al. 2023) based multi drones application.
Domaines
Robotique [cs.RO]Origine | Fichiers produits par l'(les) auteur(s) |
---|