Specifying and Verifying Active Vision-Based Robotic Systems with the SIGNAL Environment

Abstract

Active vision-based robot design involves a variety of techniques and formalisms, from kinematics to control theory, signal processing, and computer science. The programming of such systems there fore requires environments with many different functionalities, in a very integrated fashion to ensure consistency of the different parts. In significant applications, the correct specification of the global controller is not simple to achieve, as it mixes different levels of behavior, and must respect properties. In this paper we advocate the use of a strongly integrated environment that is able to deal with the design of such systems, from the specification of both continu ous and discrete parts down to the verification of dynamic behavior. The synchronous language SIGNAL is used here as a candidate in tegrated environment for the design of active vision systems. Our experiments show that SIGNAL, while not being an environment devoted to robotics (but more generally dedicated to control theory and signal processing), presentsfunctionalities and a degree ofinte gration that are relevant to the safe design of an active vision-based robotics system.