Estimator-based reactive synthesis under incomplete information

摘要: Lack of complete run-time information about the environment behavior significantly increases computational complexity and limits applicability practical reactive synthesis methods, e.g., from generalized reactivity( 1) specifications. We tackle this difficulty by splitting incomplete-information controller into estimator construction complete-information steps. The estimator, which executes in parallel to controller, establishes approximations unobserved variables that are salient for step. It essentially provides an abstraction belief space whose exponential growth often plagues synthesis, keeping track only properties relevance specification engineer scenario under consideration.We formalize notion present a framework such estimators work concert with controllers reacting partly outputs realize given temporal logic In order limit size structure estimators, we focus on positional computation. Moreover, demonstrate how well-suited be used context reactivity(1) synthesis. illustrate use estimator-based method running example motivated intelligent transportation systems.