Probabilistic reliability evaluation of autonomous navigation for firefighting robots in uncertain environments

Autonomous navigation is a mission-critical capability for firefighting robots operating in hazardous, unstructured, and highly uncertain environments. However, existing evaluation approaches are largely based on deterministic performance metrics and provide limited support for quantifying navigation reliability under uncertainty. This limitation hinders objective benchmarking and standardized assessment across different robotic platforms and operating conditions. This paper proposes an uncertainty-aware reliability assessment framework for autonomous navigation of firefighting robots. The framework integrates multi-source sensing, trajectory reconstruction, deviation analysis, and probabilistic performance evaluation into a unified pipeline. A nearest-neighbor-based trajectory matching method is employed to quantify discrepancies between reference and executed trajectories under asynchronous and non-uniform sampling conditions. To move beyond deterministic evaluation, key performance indicators, including trajectory deviation, completion time, and obstacle avoidance success rate, are modeled as stochastic variables, from which probabilistic reliability measures are derived. Experimental studies were conducted in controlled environments with static, dynamic, and complex obstacle configurations. The results show that the proposed framework effectively captures both navigation accuracy and performance variability across scenarios. In particular, the reliability metrics provide a more informative assessment than conventional threshold-based indicators by explicitly reflecting the probability of satisfying navigation requirements under uncertainty. The proposed framework offers a systematic and quantitative approach for reliability assessment of autonomous navigation systems and provides practical value for performance benchmarking, system validation, and algorithm improvement in firefighting robotics and related mobile robotic applications.