Description of LGPR System and Benefits

0440-EX-ST-2018 Text Documents

WaveSense Inc.

2018-03-20ELS_206814

 Tech Notes
  www.ll.mit.edu                                                                                                                      June 2016



Localizing Ground-Penetrating Radar
Innovative ground-penetrating radar that maps underground geological features provides
autonomous vehicles with real-time localization in all-weather conditions.

Autonomous vehicles aim to decrease
the number of driver-caused accidents
that result in over 30,000 deaths and an
estimated 2.3 million injuries in the United
States each year. Autonomous ground
vehicles (AGVs) could be deployed for
military or emergency operations in areas
that pose risks to personnel. However,
current AGVs are not mature enough for
widespread adoption. Most AGV sensors
cannot determine the vehicle’s location
when adverse conditions, such as heavy
rain or fog, snow-covered roads, or lost GPS
signals, hamper the functioning of their
sensors because those optical sensors rely
on traditional roadmap-based information
(e.g., lane markers, stop lines).               The Localizing Ground-Penetrating Radar (LGPR) uses inherently stable subsurface features
    MIT Lincoln Laboratory has developed        and their geolocation to locate the vehicle even in adverse weather conditions. The prior map
a sensor that provides real-time estimates      can be seen in gray on the left, the current scan is shown in light blue under the vehicle, and the
of a vehicle’s position even in challenging     registered data is shown in blue and green behind the vehicle.
weather and road conditions. The
Localizing Ground-Penetrating Radar             a pulse of electromagnetic radiation into            a wavelength and that it has sufficient
(LGPR) uses very-high-frequency (VHF)           the ground and measuring reflections that            contrast with the surrounding soil. The
radar reflections of underground features to    originate from scattering points below the           premise of GPR localization is that these
generate baseline maps and then matches         surface. Reflections occur at the interface          subsurface features are sufficiently unique
current GPR reflections to the baseline         between objects that have different                  and static to permit their use as identifiers
maps to estimate a vehicle’s location. The      electromagnetic properties, such as pipes,           of the precise location at which their reflec-
LGPR uses relatively deep subsurface            roots, and rocks in the surrounding “dirt.”          tions were collected.
features as points of reference because         However, it is not these discrete objects but
these features are inherently stable and less   rather the natural inhomogeneity in subter-          Mapping
susceptible to erosion or damage over time.     ranean geology that often dominates GPR              The first step in the LGPR process is to
It utilizes VHF radio waves because they        reflection profiles. Soil layers and variations      develop a map of the environment below the
can penetrate rain, fog, dust, and snow.        in moisture content cause reflections in the         road. In this step, the GPR data of subter-
                                                data. Thus, GPR paints a fairly complete             ranean “objects” are simply collected along
LGPR Methodology                                picture of the subsurface environment.               with GPS tags to form the initial database of
For subsurface sensing, GPR is one of the       With few exceptions, nearly every discrete           subsurface features. This subsurface map is
most versatile and prolific sensing modal-      object and soil feature is captured, provided        then used as a reference dataset to estimate
ities today. GPR systems work by sending        that it is not significantly smaller than            vehicle location on subsequent visits.


Tracking
Next, online localization is performed
in several steps. When the vehicle is in
motion, data are periodically fetched
from the database for matching. A
50 m × 50 m × 1–3 m three-dimensional
grid of baseline data, centered on a
GPS-defined initial location point that
is determined by the latitude, longitude,
heading, and roll (or tilt) of the sensor, is
placed in memory for matching. When the
vehicle nears the edge of this grid, it requests
a new grid of the same size centered on its
new position. In this way, a local grid of
baseline data is always maintained.
• A search region around the initial                                 The Autonomous Systems Mobile Testbed vehicle has the waterproof localizing ground-pen-
   location estimate contains “particles”                            etrating radar array mounted underneath. Additional sensors, such as lidar, camera, and
   (points on the grid) representing                                 GPS/INS units, are used for verification and studying sensor fusion.
   candidate locations and orientations.
   An algorithm iteratively evaluates                               approximately one-tenth of a center-fre-                         such a suite is its ability to operate under
   the particles to narrow the search for                           quency wavelength. This resolution is                            conditions that incapacitate other local-
   the maximum correlation within the                               finer than typically seen in GPR arrays                          ization sensors, such as optical or infrared
   vehicle’s five-dimensional space (easting,                       and enables high-fidelity matching to                            systems. Because the LGPR deduces
   northing, height, roll, and heading).                            baseline data.                                                   location on the basis of stable underground
• After several iterations, the highest-cor-                      • All elements in the array have identical                         features, it can provide position estimates
   relation particle is chosen as the most                          near-field patterns. This requirement                            even if it encounters severe weather,
   likely estimate of the vehicle’s current                         allows path retraversal to resolve                               obscured or unpaved roads, altered roads,
   location and orientation.                                        pass-to-pass offset or misalignment.                             or GPS-denied areas. Fusing LGPR with
• The search region is updated and either                                                                                            lidar or other remote sensing methods may
   expanded or shrunk to reflect this new                         Demonstrated Capability                                            provide improved localization capabilities
   estimate.                                                      The LGPR system was tested over 100s of                            for future autonomous vehicles.
                                                                  miles on paved and unpaved roads in four                               Furthermore, data from below-road
LGPR Design                                                       U.S. states and was used by the U.S. military                      features that the LGPR captures can be
The basic component of LGPR is a unique                           to navigate multiple AGVs over more than                           useful for infrastructure inspection, such as
waterproof 12-element antenna array that                          1000 miles in Afghanistan’s very demanding                         finding underground sinkholes or detecting
uses a custom VHF stepped-frequency                               environments. The LGPR method showed                               structural weaknesses in bridges. Because
continuous-wave radar. The VHF system                             robust performance in these trials.                                the LGPR compares the data it collects
penetrates deeper than typical GPR                                    The article titled “Localizing Ground                          against prior scans of the terrain, changes
systems; thus, it captures deeper, more                           Penetrating RADAR: A Step Toward Robust                            in the scans (e.g., subsurface deterioration)
stable geological features. Also, because                         Autonomous Ground Vehicle Localization,”                           can be readily detected. 
VHF frequencies are inherently insensitive                        published in the January 2016 issue of the
to small objects (e.g., a small soda can on                       Journal of Field Robotics, describes Lincoln
                                                                                                                                       Technical Point of Contact
the surface will be ignored because of its                        Laboratory’s demonstration of 4 cm cross-                            Byron Stanley
small VHF radar cross section), their use                         track localization achieved by a vehicle                             Control and Autonomous Systems
ensures that new reflections are from the                         driven at 60 mph under fair weather condi-                           Engineering Group
types of geological features cataloged in the                     tions. Recent work demonstrated real-time                            stanley@ll.mit.edu
baseline data.                                                    centimeter-level, highway-speed, nighttime                           781-981-4955
    The LGPR array of 12 dipoles is linearly                      localization during a snowstorm that had
aligned within a reflective rectangular metal                     obscured all lane markings.                                          For further information, contact
cavity with dimensions of 5 ft × 2 ft × 3 in.                                                                                          Communications and
                                                                                                                                       Community Outreach Office
Several key modifications to traditional                          Benefits of LGPR
                                                                                                                                       MIT Lincoln Laboratory
GPR were fundamental to the design:                               For an AGV to maintain awareness
                                                                                                                                       244 Wood Street
• The cavity depth of 3 inches was                                of its surroundings and location, it is
                                                                                                                                       Lexington, MA 02420-9108
   designed for under-vehicle mounting.                           equipped with a suite of sensors. The                                781-981-4204
• The spacing between array elements is                           main advantage of adding the LGPR to

   Approved for public release; distribution is unlimited.
   This material is based upon work supported under Air Force Contract No. FA8702-15-D-0001.
   Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the U.S. Air Force.

© 2016 Massachusetts Institute of Technology



Document Created: 2016-06-20 13:12:24
Document Modified: 2016-06-20 13:12:24

© 2026 FCC.report
This site is not affiliated with or endorsed by the FCC