Opacity-Preserving Task Planning for Multi-Robot Systems with LTL Specifications

Task planning remains a critical challenge in multirobot systems research. As collaborative robots increasingly operate in shared intelligent environments, they face emerging risks of privacy leakage due to the observability and inference of the execution behavior. Aiming at the security and task completion challenges of Robot planning process, we discuss how to use the current state opacity (CSO) to capture the privacy attributes of interest in robot systems to protect sensitive information. CSO ensures that external observers cannot infer confidential system states through behavioral observations. We formalize the multirobot system as modular discrete event system (DES), where each robot's behavior is modeled as a nondeterministic finite automaton (NFA). Through path redundancy reduction, we construct a secure solution space that enforces the following property: for every privacy state, there exists at least one observationally equivalent non-sensitive path that does not pass through this state. For task correctness, we encode mission requirements via linear temporal logic (LTL) formulae. We develop a synthesis algorithm that computes optimal paths on the synchronous product automaton of the LTL specifications and the secure solution space, thereby guaranteeing that all generated plans simultaneously satisfy both security and task constraints. Experimental results demonstrate the validity of the proposed framework and show that it effectively achieves task planning goals while maintaining security protection.