Machine Observations

The intention of the Machine Observations Interest Group is to discuss and document common approaches to the modelling, exchange and publication of biodiversity data derived from sensors.

GitHub

Image by Michael Librizzi
On this page

Convenor

  • Peggy Newman - Atlas of Living Australia, Australia

Core Members

  • Abigail Benson - USGS, OBIS-USA, USA
  • Holger Dettki - Swedish LifeWatch, Sweden
  • Jonathan Pye - Ocean Tracking Network, Canada
  • Peter Desmet - Research Institute for Nature and Forest (INBO), Belgium
  • Ian Jonsen - Macquarie University, Australia
  • Sarah Davidson - Max Planck Institute for Ornithology, Ohio State University, USA

Motivation

Machine observations could arguably be defined as observations that are inferred from sensor data. Sensor-based animal monitoring techniques inform a growing field of ecology and marine spatial planning that is capable of producing a very high volume of precise data about the location, internal state and environment of a diverse range of species (e.g. migratory birds, salmonids, conch, mammals both terrestrial and marine). Examples might include radio telemetry, GPS tracking, acoustic telemetry, camera traps, acoustic monitoring, or video monitoring.

For biodiversity records generated from sensors, operation and deployment information are integral as are quality control, data manipulation, and experiment design. For example information about how often a transmitter pings, or what triggers image capture, or how the species is determined, or how a sensor’s onboard algorithm determines a dive event, or how a location is calculated - provides essential context for a record. The technologies involved are in constant development, as are the design parameters of the various studies. A growing number of companies are producing smaller, longer-lasting computational devices that are fitted to a variety of animals, resulting in greater volumes of large and complex datasets. Subsequently there is a call for data standardization from all corners of the sector looking to process, analyse, document and store this valuable data.

This group is formed with three main purposes:

  1. To oversee Task Groups that document common approaches for data exchange formats for machine observations. This group was formed in the first instance to oversee a task to establish and recommend a consistent and systematic approach to applying Darwin Core (DwC) to bio-logging data because biodiversity repositories seek recommendations on the use of standards to support data publication and extraction workflows.
  2. To provide a point of contact and facilitate discussion on modelling machine observation scenarios such as biologging and camera traps and others that may arise in the TDWG community. As we work on modelling several different bio-logging scenarios, there are some obvious similarities between modelling bio-logging data and standards developed for camera traps, which also implement the same animal-sensor-deployment major entities. This group provides a point of contact and focus for examining those similarities and any others that may arise in the use of sensors for animal monitoring. Developments within the group should consider congruence with pre-existing geospatial and sensor standards.
  3. To provide a point of contact and facilitate discussion on publishing machine observations in biodiversity infrastructures. Currently these types of data tend to be housed outside the occurrence-based frameworks that have been built with human observations and specimens in focus, leaving them undiscoverable and left out of the larger biodiversity record.

Becoming involved

All practitioners are invited to share rationales for mapping their study activities into DwC or other biodiversity standards as appropriate, and to put forward their own example mappings, especially when using novel methods and/or technologies. The IG wiki will serve as an umbrella repository under which task groups can convene to produce and collate example mappings within the various domains of Machine Observations. The wiki will also provide a sounding board for the broader community to review and revise these example mappings and provide a guidepost for the community.

History and context

This group’s history is brief, just a recognition by the above group of mainly bio-loggers that there may be other groups working on similar issues in a sensor-animal-deployment paradigm.

Summary

This IG recognizes that there are growing independent communities that rely on various technologies that can be said to be producing Machine Observations. The pathways for them to contextualize their research will be as varied as their studies, but certain groups are coming together organically to seek the guidance of their peers as well as the greater biodiversity community. The set of guideposts that will emerge from the Task Groups will assist the single practitioner and the bio-logging database manager in reporting their holdings in terms that are widely understood by and useful to the biodiversity community, and the public at large.

Resources