Runtime conflict resolution mechanism for functional integrity of autonomous system

We address to a regulatory authority´s concern on a potential risk of hazard that guidance, navigation, and control(GN&C) software for autonomous systems may expose, when they operate in open, uncertain and dynamic environment. We argue an insight that the risk of hazard is derived from an unsatisfiable command of action that conflicts with systems operational limits and a prescribed safety constraints controlled by their operating environment. We replace conventional setting of GN&C problem with a scheme of computing a satisfiable series of input vectors subject to a set of constraints that represent the command of action, the safety constraints, and the operational limits. We built a coherently integrated runtime procedure comprised of a constraint solver of either computing the input vectors that realizes the command of action or detecting a subset of conflicting constraints, and a resolution mechanism of recovering satisfiabile setting. A numerical study shows that decision overhead is at most 15% of planning interval of the action, and thus our scheme is practically viable.