On Temporal Features of Airborne Odour Plumes

Animals navigate towards food, mates, and away from predators by exploiting transient fluctuations in odour concentration carried in turbulent plumes. Understanding how these fluctuations encode information about source proximity, and which strategies animals use to act on them, is essential for developing robust robotic odour source localization. This thesis presents two studies using high-resolution planar laser-induced fluorescence (PLIF) recordings of acetone vapour plumes in a benchtop wind tunnel (Connor, McHugh and J. P. Crimaldi, 2018), spanning four release configurations: near-bed release at 10 cm/s, and freestream release at 5 cm/s, 10 cm/s, and 20 cm/s. The first study extends the bout analysis pipeline of Schmuker, Bahr and Huerta (2016), originally developed for metal oxide (MOX) gas sensors, to optically recorded PLIF plume data. Bout count, duration, amplitude, and inter-bout interval (IBI) each exhibit systematic spatial structure: bout counts follow a non-linear sigmoidal decline with downwind distance along the plume centre line, crosswind profiles are approximately Gaussian, and IBI increases with source distance as an inverted Gaussian. Principal component analysis confirms that these statistics together encode odour source proximity. The second study applies three biologically inspired in-silico navigation strategies to the same PLIF data: a direction-sensing (DS+) and non-direction-sensing (DS−) strategy modelled on walking Drosophila, and a cast-and-surge (CS) strategy modelled on flying insects. The CS strategy outperformed both walking strategies across the majority of release configurations. The dominant factor was the behaviour adopted in the absence of odour: crosswind casting was consistently more effective than random walking at relocating the plume. Together, these findings establish a quantitative link between turbulent plume structure and the transient odour statistics available to navigating animals and robots.