Policy Synthesis for Metric Interval Temporal Logic with Probabilistic Distributions

Abstract: Metric Temporal Logic can express temporally evolving properties with time-critical constraints or time-triggered constraints for real-time systems. This paper extends the Metric Interval Temporal Logic with a distribution eventuality operator to express time-sensitive missions for a system interacting with a dynamic, probabilistic environment. This formalism enables us to describe the probabilistic occurrences of random external events as part of the task specification and event-triggered temporal constraints for the intended system's behavior. The main contributions of this paper are two folds: First, we propose a procedure to translate a specification into a stochastic timed automaton. Second, we develop an approximate-optimal probabilistic planning problem for synthesizing the control policy that maximizes the probability for the planning agent to achieve the task, provided that the external events satisfy the specification. The planning algorithm employs a truncation in the clocks for the timed automaton to reduce the planning in a countably infinite state space to a finite state space with a bounded error guarantee. We illustrate the method with a robot motion planning example.