Spatio-Temporal Planning for a Reconfigurable Multi-Robot System
Thomas M. Roehr, Frank Kirchner
Editors: Alberto Finzi, Erez Karpas
In Proceedings of the 4th Workshop on Planning and Robotics, (PlanRob-2016), 13.6.-14.6.2016, London, o.A., pages 135-146, Jun/2016.
In this paper we introduce a spatio-temporal planning and scheduling approach for collaborative multi-robot systems. In particular, we are targeting an application to physically reconfigurable systems in order to take advantage of morphological changes. The planning approach relies on an ontology to model the functionalities individual physical agents offer within the multi-robot system, while an implicit domain representation is given. An inference layer on top of a knowledge-based system allows to account for superadditive effects from physically combining two or more robots. We present a formulation of the domain-specific planning problem and outline our spatio-temporal planning approach. This approach combines the use of constrained-based satisfaction techniques with linear optimization to solve a multicommodity min-cost flow problem to deal with the transportation of immobile robotic systems. The paper describes the findings after implementing core features and evaluates the approach based on a fictitious scientific mission. We close with a discussion of the current limitations of the illustrated approach.