Synthesis of Surveillance Strategies via Belief Abstraction

Abstract

We study the problem of synthesizing a controller for a robot with a surveillance objective, that is, the robot is required to maintain knowledge of the location of a moving, possibly adversarial target. We formulate this problem as a one-sided partial-information game in which the winning condition for the agent is specified as a temporal logic formula. The specification formalizes the surveillance requirement given by the user, including additional non-surveillance tasks. In order to synthesize a surveillance strategy that meets the specification, we transform the partial-information game into a perfect-information one, using abstraction to mitigate the exponential blow-up typically incurred by such transformations. This enables the use of off-the-shelf tools for reactive synthesis. We use counterexample-guided refinement to automatically achieve abstraction precision that is sufficient to synthesize a surveillance strategy. We evaluate the proposed method on two case-studies, demonstrating its applicability to large state-spaces and diverse requirements.

Strategy Synthesis for Surveillance-Evasion Games with Learning-Enabled Visibility Optimization

“…We note that to avoid the state space blow-up arising from the subset construction in the partial information game, we solve the surveillance games using the belief-set abstraction techniques detailed in [22].…”

“…Synthesizing a strategy for a surveillance specification is inherently reactive, the agent observes the target's action and reacts appropriately in order to satisfy its specification. However, planning only the immediate action, as is done in [22], will not result in a large gain in visibility coverage.…”

Distributed Synthesis of Surveillance Strategies for Mobile Sensors

“…We then solve each game individually with local specifications local i ( p 5 ). We solve these singleagent surveillance games using an abstraction-based method detailed in a companion publication at CDC 2018, detailed in [16]. We report the synthesis times in Table I.…”

“…In a companion publication at CDC 2018 we describe a framework for formalizing single-agent surveillance synthesis as a two-player game with partial information, and propose an abstraction-based method for solving such games. The interested reader is referred to the extended version [16] for details about the abstraction-based synthesis method. The price of resorting to abstraction is the potential overapproximation of the set of possible target locations (that is, loss of precision in the sensors' knowledge) which may make satisfying the surveillance requirements more difficult.…”

Privacy Verification and Enforcement via Belief Abstraction

“…By proving that the belief dynamics is mixed monotone, we could efficiently obtain the abstracted finite state system that serves as an overapproximation of the underlying continuous dynamic [15], which has been successfully applied on the traffic network control with temporal logic specifications [16]. The belief abstraction idea has also been proposed in [17], but their belief space is the power set of the state space, which is discrete and finite.…”