No abstract

This paper addresses the problem of estimating a likelihood map for the location of the source of a chemical plume using an autonomous vehicle as a sensor probe in a fluid flow. The fluid flow is assumed to have a high Reynolds number. Therefore, the dispersion of the chemical is dominated by turbulence, resulting in an intermittent chemical signal. The vehicle is capable of detecting above-threshold chemical concentration and sensing the fluid flow velocity at the vehicle location. This paper reviews instances of biological plume tracing and reviews previous strategies for a vehicle-based plume tracing. The main contribution is a new source-likelihood mapping approach based on Bayesian inference methods. Using this Bayesian methodology, the source-likelihood map is propagated through time and updated in response to both detection and nondetection events. Examples are included that use data from in-water testing to compare the mapping approach derived herein with the map derived using a previously existing technique.

This paper addresses the problem of mapping likely locations of a chemical source using an autonomous vehicle operating in a fluid flow. The paper reviews biological plume-tracing concepts, reviews previous strategies for vehicle-based plume tracing, and presents a new plume mapping approach based on hidden Markov methods (HMM). HMM provide efficient algorithms for predicting the likelihood of odor detection versus position, the likelihood of source location versus position, the most likely path taken by the odor to a given location, and the path between two points most likely to result in odor detection. All four are useful for solving the odor source localization problem using an autonomous vehicle. The vehicle is assumed to be capable of detecting above threshold chemical concentration and sensing the fluid flow velocity at the vehicle location. The fluid flow is assumed to vary with space and time, and to have a high Reynolds number (Re>10).

a This paper describes an optofluidic droplet interrogation device capable of counting fluorescent drops at a throughput of 254 000 drops per second. To our knowledge, this rate is the highest interrogation rate published thus far. Our device consists of 16 parallel microfluidic channels bonded directly to a filtercoated two-dimensional Complementary Metal-Oxide-Semiconductor (CMOS) sensor array. Fluorescence signals emitted from the drops are collected by the sensor that forms the bottom of the channel. The proximity of the drops to the sensor facilitates efficient collection of fluorescence emission from the drops, and overcomes the trade-off between light collection efficiency and field of view in conventional microscopy. The interrogation rate of our device is currently limited by the acquisition speed of CMOS sensor, and is expected to increase further as high-speed sensors become increasingly available.