Information-Driven Active Perception for k-step Predictive Safety Monitoring

Abstract

This work studies the synthesis of active perception policies for predictive safety monitoring in partially observable stochastic systems. Operating under strict sensing and communication budgets, the proposed monitor dynamically schedules sensor queries to maximize information gain about the safety of future states. The underlying stochastic dynamics are captured by a labeled hidden Markov model (HMM), with safety requirements defined by a deterministic finite automaton (DFA). To enable active information acquisition, we introduce minimizing k-step Shannon conditional entropy of the safety of future states as a planning objective, under the constraint of a limited sensor query budget. Using observable operators, we derive an efficient algorithm to compute the k-step conditional entropy and analyze key properties of the conditional entropy gradient with respect to policy parameters. We validate the effectiveness of the method for predictive safety monitoring through a dynamic congestion game example.

Figures (6)

• Figure 1: Environment topological graph with sensor coverages.
• Figure 2: State transition graph representing the ego agent's goal policy.
• Figure 3: Failure dfa for the dynamic congestion game example.
• Figure 4: Comparison of $k$-step prediction accuracy with posterior sampling.
• Figure 5: Convergence of the policy gradient method for $k=3$, $\alpha=0.04$.

Theorems & Definitions (18)

• Definition 1: Labeled-hmm
• Definition 2: ltlf de2013linear
• Definition 3: dfa
• Definition 4: Product hmm
• Definition 5: State-based $k$-step predictability
• Definition 6: Observable operator jaeger2000observable
• Proposition 1
• Proposition 2
• Proposition 3
• Proposition 4