Multi-robot autonomous search for radioactive leakage source based on a CPE-Infotaxis method during nuclear emergency rescue

In nuclear emergency response, minimizing radiation exposure is critical, yet single robots struggle to efficiently locate leakage sources in complex turbulent environments. Multi-robot Infotaxis offers a promising solution but remains vulnerable to radiation-induced sensor failures that introduce false information and corrupt shared beliefs. To address these challenges, we propose the CPE-Infotaxis (Cognitive Probabilistic Entropy Infotaxis) algorithm, in which each robot dynamically adjusts the credibility of received information through relative entropy-weighted social Bayesian estimation, and executes a distributed Infotaxis search strategy based on this assessment, while integrating an exploration-exploitation balancing strategy and an adaptive step-size mechanism to mitigate excessive exploration. The proposed CPE-Infotaxis method is evaluated against both the traditional Infotaxis and PE-Infotaxis algorithms under ideal and fault-detection scenarios. Experimental results show that CPE-Infotaxis improves search efficiency by 40.7% compared to traditional Infotaxis and by 38.7% compared to PE-Infotaxis(Exploration Probability Infotaxis), providing robust technical support for autonomous radioactive leakage source search in nuclear emergency response.