Systems of Systems (SoS) in critical domains like construction require the coordination of independent and heterogeneous Constituent Systems (CS) to accomplish complex missions. To help with such coordination, an architectural approach, called orchestration, has been proposed. However, safety in such an approach remains unexplored. In this paper, we present a safety-aware strategy synthesis framework to fill this gap. It combines formal modeling of CS and shared resources as timed automata, integration of safety contracts to capture assumptions and guarantees, and Q-learning strategy generation by using Uppaal Stratego. As a result, the framework enables the synthesis of execution strategies that not only fulfill mission objectives but also ensure safety constraints. We demonstrate our method through a case study in autonomous construction operations, highlighting its ability to minimize unsafe interactions and to reduce resource conflicts and waiting times.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Safety-Aware Strategy Synthesis for Autonomous System of Systems with UPPAAL

  • Nazakat Ali,
  • Muhammad Naeem,
  • Julieth Patricia Castellanos Ardila,
  • Sasikumar Punnekkat

摘要

Systems of Systems (SoS) in critical domains like construction require the coordination of independent and heterogeneous Constituent Systems (CS) to accomplish complex missions. To help with such coordination, an architectural approach, called orchestration, has been proposed. However, safety in such an approach remains unexplored. In this paper, we present a safety-aware strategy synthesis framework to fill this gap. It combines formal modeling of CS and shared resources as timed automata, integration of safety contracts to capture assumptions and guarantees, and Q-learning strategy generation by using Uppaal Stratego. As a result, the framework enables the synthesis of execution strategies that not only fulfill mission objectives but also ensure safety constraints. We demonstrate our method through a case study in autonomous construction operations, highlighting its ability to minimize unsafe interactions and to reduce resource conflicts and waiting times.