Where EEs Navigate A Changing World.
Globe

Sensors in the lab and in the field

by Mike Stanley

%caption
Figure 1: Norway Rat

I recently came across two technical papers which piqued my curiosity.† Both used wireless sensors to extend scientists ability to gather data from lab animals. Wireless Inertial Sensors for Monitoring Animal Behavior [1] explores the use of wireless sensors to monitor behavior of caged laboratory animals.† 3-axis accelerometers were attached to lab rats using specially designed "rat jackets".† Collected data was fed into a series of neural nets, which were then trained to recognize various "rat behaviors".† Power consumption for the smart jacket was minimal.† Researchers estimated that a sample rate of 36 samples/second can be sustained for over five days.

Results were impressive.† As you would expect, itís easy to identify when a rat is sleeping:† accelerometer readings vary slowly.† The neural nets were also able identify when rats were grooming or eating (both around 93% accuracy) or standing (97% accuracy).†

At first, this might seem like an academic exercise, until you realize that millions of animals are used for laboratory experimentation every year.† Monitoring behavior wirelessly has the potential to not only provide valuable new data for researchers, but also increase the well-being of lab animals by using less intrusive, and yet more thorough, monitoring of behavior.

The second paper has a title that forces you to sit down for a casual read. "BurrowView - Seeing the world through the eyes of rats" [2]† (what geek could resist that?) takes the concept from laboratory to the field.† This time, our rat buddies were outfitted with special backpacks which included 3 axis accelerometer, 2 axis gyro, 3 axis compass, microcontroller and radio.†† Researchers borrowed from work in pedestrian navigation to correlate sensor readings with the gait of the animals.†† They were thus able to extract orientation and position over time.† This allowed generation of "pathlets", each of which represented a single ratís movement over some short period of time.

Specialized algorithms borrowing from SLAM (Simultaneous localization and mapping) techniques were to merge multiple pathlets into an overall view of the rat burrow.† The description of how this was accomplished, by itself, is worth your investment in reading the paper.† Researchers used a technique called "sphere filtering" to sort through sensor noise, varying rat behavior and redundant pathlets to generate the overall map.

Although both papers nominally deal with the rodent world, I think youíll immediately see parallels in other, more human, activities.†† If youíre aware of similar work that you would like to share, please send an email.† We would love to hear from you.

Footnotes:

  1. Proceedings of the 29th Annual International Conference of the IEEE EMBS, Subramaniam Venkatraman, John D. Long, Kristofer S.J. Pister and Jose M. Carmena, 2007.
  2. The Second International Workshop on Information Quality and Quality of Service for Pervasive Computing,† Jo Agila Bitsch Link, Gregor Fabritius, Muhammad Hamad Alizai, Klaus Wehrle, 2010.