Publication Detail

Development and application of distributed MEMS pressure sensor array for AUV object avoidance

Vicente I. Fernandez, Stephen M. Hou, Franz S. Hover
9 pp.
MITSG 09-07
$5.50 DOM / $7.50 INT

A novel sensory system is being developed for AUVs
to augment current sensory systems for navigation and
operation in difficult environments. These environments
are frequently cluttered and murky with substantial flow
from currents or waves, frustrating sonar and vision systems
while posing an increased risk to AUVs. In order
to manage such situations, a better ability to locate
and identify physical objects is needed. This gap could
be filled by small low frequency pressure sensors distributed
over the surface of the AUV in dense arrays.
The pressure sensor array presented here consists of
hundreds of MEMS pressure sensors with diameters
near 1 mm spaced a few millimeters apart fabricated
on etched silicon and Pyrex wafers; a fabrication process
for producing the array is described. A strain-gauge
pressure sensor is analyzed, fabricated, and tested. It satisfies
specifications as required for object detection. The
sensing element is a strain gauge mounted on a flexible
1This publication is the result of research sponsored by The
MIT Sea Grant College Program, under NOAA grant number
NA06OAR4170019, project number R/RT-2/RCM-17.
diaphragm, which is a thin (20 μm) layer of silicon attached
at the edges to a square silicon cavity 2000 μm
wide on a side. A source voltage of 10 V produces a
sensor with a sensitivity on the order of 1 μV/Pa. Since
the thermal noise voltage is near 0.7 μV, the pressure
resolution of the sensors is on the order of 1 Pa.
In addition to a pressure sensor array capable of measuring
the spatial pressure distribution, progress has also
been made in estimating the shape of an arbitrary twodimensional
physical object in a flow and in tracking
vortices based solely on distributed pressure measurements.
The shape estimation relies on a conformal mapping
which orders shape parameters by their observability
with range. Utilizing this model, a sequential maximum
likelihood estimation technique is able to extract
the contour of an object in steady flow. This procedure
requires no a priori knowledge of the type of object.
The result is an estimate of a completely arbitrary shape
whose level of generality depends on the distance of the

type: Technical reports

Parent Project

Project No.: 2006-R/RT-2/RCM-17
Title: Touch-at-a-Distance: Pressure Microsensor Arrays for AUV Navigation

Other publications by this author (as Lead)