M O N a D I C R O B O T I C S Monadic Robotics

We have developed a domain speci c language for the construction of robot controllers, Frob (Functional ROBotics). The semantic basis for Frob is Functional Reactive Programming, or simply FRP, a purely functional model of continuous time, interactive systems. FRP is built around two basic abstractions: behaviors, values de ned continuously in time, and events, discrete occurances in time. On this foundation, we have constructed abstractions speci c to the domain of robotics. Frob adds yet another abstraction: the task, a basic unit of work de ned by a continuous behavior and a terminating event. This paper examines two interrelated aspects of Frob. First, we study the design of systems based on FRP and how abstractions de ned using FRP can capture essential domain-speci c concepts for systems involving interaction over time. Second, we demonstrate an application of monads, used here to implement Frob tasks. By placing our task abstraction in a monadic framework, we are able to organize task semantics in a modular way, allowing new capabilities to be added without pervasive changes to the system. We present several robot control algorithms specied using Frob. These programs are clear, succinct, and modular, demonstrating the power of our approach.