ALARM eResearch Australasia 2012

Presented at eResearch Australasia.

Transcript

1. ALARM Richard Sinnott, Gerson Galang, Stephen Marshall and Christopher Bayliss
2. Aims —  Monitor water quality —  Directly via sensors. —  Indirectly via animals. —  In real time or as close as possible. —  Correlate with catchment information. To locate sources of pollution.
3. Current methods —  People in field taking samples —  People standing in rivers with sensors and nets. —  Automated samplers —  Grab and freeze a sample regularly. —  Someone collects it for lab analysis.
4. Animal monitoring —  An animal reacts to its environment. —  By monitoring the behavior of an animal in the water we can deduce things about the water quality. —  Previous approaches —  Indicator species —  Tank
5. Behaviors —  Movement —  Gill movement —  Feeding —  Hiding
6. Rough Outline Sensors NBN Collector Analytics Website
7. Sensors —  Cheap —  Low power —  Can handle multiple sensors
8. Sensor Hardware —  BeagleBoard —  Arduino —  USB camera —  G4 Modem
9. Sensor Software —  Linux —  Python —  Gstreamer —  WebM video codec —  OpenCV vision library —  Arduino flavour of C
10. Collector —  Receives data from sensors —  Stores raw data —  Converts raw data into usable values. —  Uses python and PostgreSQL with PostGIS
11. Website —  Java and JavaScript based —  Allows users to search and visualise collected data.
12. Current status —  Lab based test bed for controlled testing. —  Laptop based test sensor. —  Building a library of video and sensor data. —  Prototype sensor package. —  Temperature, pH, Redox … —  Experimenting with WAN compatible data transport options. AMQP , XMPP , … —  Developing video analysis filters. —  Migrating from Matlab to Python + OpenCV
13. Demo