Adaptive Robotic Source Seeking in Vortical Indoor Environments With Sparse Structural Guidance

Indoor source seeking is vital for tasks such as security monitoring and disaster rescue, yet it remains highly challenging in environments characterized by turbulent airflow and sparse structural guidance. In such settings, airflow directions may shift abruptly due to obstacle-induced vortices, while odor cues are often intermittent and spatially sparse. These factors increase the difficulty of crosswind search in sparse space. This paper proposes a perception-guided adaptive search algorithm that fuses short-term odor and wind observations to assess the reliability of perceptual cues and adaptively adjust the robot's search behavior. By refining the search region based on perceptual cues, the algorithm effectively regulates crosswind motion, thereby enabling timely escape from vortex-induced traps and improving source search efficiency. The proposed algorithm is evaluated across diverse indoor environments and inlet-outlet configurations under different airflow speeds. It is further compared with the representative state-of-the-art algorithms and shows superior performance in simulation experiments. Real-robot experiments are also conducted to verify its practical effectiveness in physical indoor environments.