A runtime system for logical-space programming

Abstract

In this paper we introduce logical-space programming, a spatial computing paradigm where programs have access to a logical space model, i.e., names and explicit relations over such names, while the runtime system is in charge of manipulating the physical space. Mobile devices such as autonomous vehicles are equipped with sensors and actuators that provide means for computation to react upon spatial information and produce effects over the environment. The spatial behavior of these systems is commonly specified at the physical level, e.g., GPS coordinates. This puts the responsibility for the correct specification of spatial behaviors in the hands of the programmer. We propose a new paradigm named logical-space programming, where the programmer specifies the spatial behavior at a logical level while the runtime system is in charge of managing the physical behaviors. We provide a brief explanation of the logical-space computing semantics and describe a logical-space runtime system using bigraphs as logical models and bigActors as computing entities. The physical entities are modeled as polygons in a geometrical space. We demonstrate the use of logical-space programming for specifying and controlling the spatial behaviors of vehicles and sensors performing an environmental monitoring mission. The field test consisted of an Unmanned Aerial Vehicle and GPS drifters used to survey an area supposedly affected by illegal bilge dumping.