This project tackles the problem of robotic odour source localisation, that is, the use of robots to find the source of a chemical release. As the odour travels away from the source, in the form of a plume carried by the wind or current, small scale turbulence causes it to separate into intermittent patches, suppressing any gradients and making this a particularly challenging search problem. We focus on distributed strategies for odour plume tracing in the air and in the water and look primarily at 2D scenarios, although results are also presented for 3D tracing.
The common thread to our work is the use of multiple robots in formation, each outfitted with odour and flow sensing devices. By having more than one robot, we can gather observations at different locations, thus helping overcome the difficulties posed by the patchiness of the odour concentration. The flow (wind or current) direction is used to orient the formation and move the robots up-flow, while the measured concentrations are used to centre the robots in the plume and scale the formation to trace its limits.
Embed of video is only possible from Mediaspace, SwitchTube, Vimeo or Youtube