Recursive and Backward Reasoning in the Verification on Hybrid Systems

In this paper we introduce two improvements to the method of verification of hybrid systems by constraint propagation based abstraction refinement that we introduced earlier. The first improvement improves the recursive propagation of reachability information over the regions constituting the abstraction, and the second improvement reasons backward from the set of unsafe states, instead of forward from the safe of initial states. Detailed computational experiments document the usefulness of these improvements.