Simulation of visual sensory information in an advanced robot programming environment

The off-line simulation of robot tasks will play an increasingly important role in the development of robot applications, particularly as they diversify into more complex and unstructured domains. In these domains the availability of good sensory information becomes vital to the success of the task, off-line simulation must therefore simulate not only the robot movements but also the sensory information on which these movements are based. One of the most powerful sensory mechanisms available to a robot is vision but little has been done so far to incorporate visual sensory simulation into robot simulation. This paper describes a method for generating accurate visual sensory information during the simulation of a robot task and shows how this can enable the complete simulation of the task, particularly the connection between sensing and action.<>