<p>Accurate localization of hazardous gas leaks, like 2-heptanone, is crucial for chemical environment safety. Conventional devices struggle to differentiate similar compounds and ignore airflow effects. This study demonstrates that modifying the interfacial and gas-phase reactivity of metal oxides such as In₂O₃ enhances gas specificity. Density functional theory (DFT) calculations reveal that In₂O₃ (222) surfaces strongly adsorb and interact electrochemically with 2-heptanone, boosting interfacial catalytic activity. Gas-phase conversion tests show that In₂O₃ calcined below 1000 °C reduces interference from other gases, improving selectivity for 2-heptanone. The optimized sensor, using an In₂O₃ electrode, exhibits high sensitivity and selectivity for 2-heptanone. To locate leaks accurately, we developed a wind-aware robot equipped with this sensor. The robot counters airflow effects using light detection and ranging (LiDAR) for spatial mapping and ultrasonic anemometry for wind measurement. By integrating real-time wind-field modeling with a Twin Delayed Deep Deterministic Policy Gradient (TD3) reinforcement learning algorithm, the robot can interpret chemical signals under fluctuating wind conditions, enabling precise leak localization. Simulations and wind tunnel tests yield localization success rates of 90% and 70%, respectively. This work combines advanced material design, aerodynamic intelligence, and autonomous navigation, providing a practical solution for gas leak detection in complex environments.</p> Graphical abstract <p></p>