Iterative Controller Synthesis for Multirobot System

The synthesis problem is to construct a system fulfilling some specific requirements when interacting with the environment, which is one of the most crucial and challenging tasks in robotics. In comparison to the case of dealing with a single robot, synthesizing of a system constituted with multiple robots is, in general, much more involved. Actually, information is shared among robots in the latter case, and for a fixed robot, when fictively merging the rest ones into its environment, we are confronted with imperfect description of environments. In this article, we present an iterative controller synthesis approach to dealing with multirobot systems. In our model, the behaviors and outputs of one robot can be observed by the other ones, as a part of their inputs. To make our model more flexible, we allow the mechanism of “partial observation,” namely, only a part of outputs can be observed by other robots on some particular sensors. Our synthesis approach is conductive in an iterative manner. By analyzing the dependence among robots, we first try to synthesize controllers for some of them and then extract a set of invariants from the solved part to refine other ones. Repeatedly and iteratively using this way, we may arrive at a complete solution. Meanwhile, in comparison to the monolithic approach, using the iterative manner usually produces a much more compact result, which means that the size of the controller is smaller.