\ I /
United States
Environmental Protection
Agency
Risk Reduction
Engineering Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-89/055 Feb. 1990
Project Summary
Demonstration of
Autonomous Air
Monitoring Through Robotics
Robert J. Rancatore and Michael L. Philips
Hazardous and/or tedious functions
are often performed by on-slte
workers during Investigation, mitiga-
tion and clean-up of hazardous
substances. These functions include
site surveys, sampling and analysis,
excavation, and treatment and prepa-
ration of wastes for shipment to
chemical disposal sites. Many times,
people working in the "hot zone"
must leave their primary work on a
frequent basis to perform the tedious
task of monitoring the air at the
perimeter of the clean-up site. The
project objective was to demonstrate
the potential usefulness of air sam-
pling (monitoring) utilizing a commer-
cially available, field compatible robot
with minimum modification.
This project included modifying an
existing teleoperated robot to include
autonomous navigation, large object
avoidance, and air monitoring and
demonstrating that prototype robot
system in indoor and outdoor envi-
ronments. The prototype robot sys-
tem successfully performed each of
its required tasks.
This Project Summary was devel-
oped by EPA's Risk Reduction Engi-
neering Laboratory, Cincinnati, OH, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Investigation, mitigation, and clean-up
of hazardous substances often require
on-site workers to perform hazardous
and/or tedious functions. These functions
include site surveys, sampling and
analysis, excavation, and treatment and
preparation of wastes for shipment to
chemical disposal sites. Workers per-
forming these functions risk dermal,
ocular, or inhalation exposure to haz-
ardous chemicals. At many clean-up
sites, people working in the "hot zone"
will regularly be taken away from their
primary work each hour to perform the
tedious task of monitoring the air at the
perimeter of the clean-up site. Having an
autonomous device reliably perform
routine monitoring tasks at clean-up sites
would be effective, particularly a device
requiring minimum effort to setup and
control.
Robotic devices have been developed
and used for a variety of functions
including use in hazardous environments,
such as those in nuclear power plants.
Using robotic devices at Superfund/SARA
operations can possibly increase worker
safety by removing the worker from the
potential for chemical contact or expo-
sure to fire, explosion, or other physical
injury. Robotic devices also offer the
potential of more rapid and reliable
performance of those tedious tasks which
may become hazardous if the tedium
causes worker carelessness. This project
was undertaken to demonstrate the use
of a commercially available robotic plat-
form for air monitoring during removal or
remedial activities.
The tests performed in the demon-
stration are described in the Performance
Demonstration Plan titled "Air Monitoring
Autonomous Robot System Demon-
stration Plan" submitted and approved in
December 1988.
-------
Approach
To demonstrate an autonomous air
monitoring robot, an existing teleoperated
Surveyor®* robot (Figure 1), developed
by ARD Corporation was modified to an
autonomously navigated robot. A mag-
netic pick-up unit was installed on the
Surveyor® so that the Surveyor® could
respond to and correct its course in
relation to a high pressure electric pulse
travelling along a wire laid on the ground.
The robot was also modified to carry an
HNU PI-101 Photoionization Detector air
monitoring device. A sonar range finder,
which already was an integral part of the
Surveyor®, was repositioned to the front
of the robot chassis to detect large
obstacles in the path of the robot, thereby
bringing the robot to a stop.
The software of the onboard computer
was extensively modified to provide
autonomous navigation control, dynamic
steering to smoothly follow the wire-
course without hesitation, obstacle avoid-
"Mention of trade names or commercial products
does not constitute endorsement or
recommendation for use.
Figure 1. ARD Surveyor® robot and command console.
ance, autonomous shut down and remol
reporting of toxic substance detection.
addition to the electrical interfaces, phy:
ical mounts constructed for the hardwai
were used to attach the sensors to pr<
viously existing mount points on th
robot. Thus the system's water tight inti
grity was not compromised. Surveyor®
rated as water tight to 10 ft, although th
practical limitation of the height of th
communications antennas prevents i
use in more than 14 in. of water.
Robot System Testing
Developmental testing was conduct*
at the ARD's facilities in Columbi
Maryland, and indoor and outdoor tes
were conducted at the EPA site
Edison, New Jersey. Incremental pro
lems were resolved before the next tes
were conducted.
Four trials were run before the demo
stration to ensure that the systems we
correctly modified for guide wire dete
tion, wire following, obstacle negotiatic
obstacle avoidance, terrain negotiatic
and toxic substance detection.
Outdoor Demonstration,
Edison, New Jersey
Approach
The outdoor demonstration of tl
robot system was held in an open ar
overgrown with 2- to 3-ft grass, weec
and small trees. The uneven surface w
small gullies was representative of t
type of terrain found at an actual was
site. The guide wire was placed on t
ground in a kidney shaped form with t
perimeter somewhat wavy and uneven.
A simulated toxic substance (pa
thinner) was placed in aluminum pa
alongside the wire so that the robi
mounted air monitor could detect t
substance as it passed by. The air me
itor continuously updated the video d
play at the control console with t
number of parts per million of to:
substance that it found in the air as w
as with a message that said it located 1
substance whenever vapors were c
tected. This was recorded on video ta
for documentation and analysis. At 1
same time, a technician walking besi
the robot with a hand held air moni
verbally announced his monitor's re<
ings. His verbal reports were picked
by the onboard microphone a
recorded with the video image tal<
through the drive cameras of the rot
In addition to these video taped portic
of the demonstration, a separate V
Camcorder recorded the proceedir
demonstration from approximately 151
away from the demonstration site.
-------
'esufts
The robot was teleoperatively maneu-
ered onto the guide wire and then
ilaced in autonomous mode by a switch
m the control console. From that point
m, the robot controlled its own navigation
md negotiated the course in approxi-
nately 15 min. Along the way, the robot
vas required to navigate over small
ibstacles, record the concentration of the
oxic substance and stop itself if a large
ibstacle was detected. The robot per-
ormed all of these tasks successfully
md followed the guide wire without
turnan assistance. Concern that sharp
urns in the course would be a problem
or the navigation system did not mate-
ialize.
During the demonstration, the onboard
;amera was active and the video and
ludio was transmitted from the robot to
he command console where it was
Jisplayed on the monitor and recorded
jy a VHS Recorder. Superimposed on
his same image was the concentration
evel of the toxic substance as reported
3y the air monitor The autonomous and
nanually read outdoor air monitor read-
ngs can be compared (Table 1). The
3ans of paint thinner were placed at two
different locations; therefore, there are
wo sets of comparisons.
able 1. Comparison of Outdoor Man-
ual and Autonomous Air Moni-
tor Sensor Readings, ppm
Data
Manual
0
16
20
20
7
20
10
3
0
Set if 1
Robot
0
19
19
19
10
10
3
2
0
Data
Manual
0
10
10
6
7
5
0
—
—
Set #2
Robot
0
0
0
0
0
0
0
—
—
The first set of data taken by the
onboard air monitor compared very well
with that of'the manual air monitor.
Although larger differences were ex-
pected because the air monitors were
different and because they were held in
slightly different locations, we, instead,
received very good agreement with an
RMS error of +/- 22%. This result is all
the more remarkable given the fact that
there was a very strong wind that was
estimated to be blowing above 25 mph.
In the second set of outdoors data, the
system did not detect any "toxic" sub-
stance. Note, however, that the maximum
manual readings were only half those of
the first data set and that the overall
readings were much lower. These
changes are believed primarily due to the
high wind and the direction of the wind.
Wind direction was a factor because in
certain orientations of the robot as it
travelled around the course, e.g., where
the first set of data was recorded, the
robot blocked the wind and kept it from
affecting the readings of both air mon-
itors. During the second set of readings,
however, the air monitor onboard the
robot was not shielded from the wind.
The manually held monitor, however, was
partially shielded from the wind by the
operator resulting in the manually held
monitor detecting a small amount of
"toxic" substance while the robot-
mounted monitor did not detect any of
the substance.
Indoor Demonstration
Approach
The indoor demonstration was similar
in concept to the outdoor demonstration
and took place in a large warehouse. A
course similar to the one outdoors,
though smaller, was placed on the flat
concrete floor. The robot system was
once again maneuvered over the guide
wire in teleoperation mode, placed in
autonomous mode, and allowed to follow
the guide wire. Aluminum pans of paint
thinner were placed beside the wire at
two locations so that the robot-mounted
air monitor could detect the "toxic" sub-
stance as it passed by the pans. Again, a
technician walking beside the robot with a
hand held air monitor took and verbally
announced these readings. His verbally
announced readings and the robot's
readings were recorded on the video tape
for later analysis. The remainder of the
tasks that were demonstrated outdoors
were repeated indoors. The whole indoor
demonstration was recorded on video
tape.
Results
With the exception of one test, all of
the tasks were performed successfully:
autonomous navigation, small obstacle
navigation, large obstacle detection, and
accurate reporting of the detection of a
toxic substance. In one of the small
obstacle navigation tests, the robot lost
the signal from the wire and came to a
stop after navigating over one of the 2 x 4
planks that was placed across the guide
wire. This isolated incident may have
been caused by a loose connection in
one of the guide wire sensors.
In Table 2, the autonomous and manual
air monitor readings can be compared.
Two sets of data were taken at different
locations around the course. The first set
of data was recorded directly in front of
an open door so that the wind affected
the readings of the air monitor. As can be
seen from Table 2, each set of readings
compares very well with the minor
exception that the first set of onboard
readings continue to detect a toxic sub-
stance in the air for a short time after the
manually held readings had returned to
background levels. The data yields an
RMS error of +/- 18%. The robot system
was travelling downwind of the pans of
the "toxic" substance, and we believe the
wind carried the fumes to the air monitor
where it continued to detect a low
concentration level of the substance. The
onboard system and the manual system
reached the same maximum value and
began detecting the substance at almost
the same time.
Table 2. Comparison of Indoor Manual
and Autonomous Air Monitor
Sensor Readings, ppm
Data
Manual
0
20
20
20
10
0
0
0
0
Set if 1
Robot
0
19
19
19
19
5
2
2
0
Data
Manual
0
20
20
20
20
20
5
3
0
Set #2
Robot
0
19
19
19
19
19
5
2
0
The second set of data were taken with
the pans of "toxic" substance much
farther away from the door. The readings
match extremely well with the beginning
of detection, the ending of detection, and
the rise and fall of the levels of con-
centration. The RMS error for this data is
+ /-3%.
In addition to repeating the tasks per-
formed outdoors, the robot was placed in
teleoperation mode, and its ability to
climb and descend a flight of stairs was
demonstrated. The optics capability of
the robot was also demonstrated using
the Charge-Couple Device (CCD) cam-
era. Pan and zoom were demonstrated
with the image displayed on the com-
mand console.
-------
Conclusions and
Recommendations
An autonomously guided robot suc-
cessfully performed air monitoring of
"toxic" substances both indoors and out-
doors. An existing robot platform could
be modified to perform tedious yet
important tasks by adding specific sen-
sors to the robot and tying them into the
communications system of the robot.
While demonstrating the autonomous
air monitoring system, several areas were
discovered where the existing robot plat-
form should be further modified before it
can be implemented as a field robot. To
improve the Surveyor's utility in the field,
the following modifications are recom-
mended:
• investigate and develop an active
suspension for all terrain use;
• change drive motors, increase motor
speed, and add reduction gears to
increase vehicle speed from 1 fps
(foot per second) to an operator
specified 1 to 10 fps;
upgrade the onboard micro-processor
to an 80286 or 80386 class machine,
depending on processing require-
ments;
modify software to relocate the guide
wire and to continue on course if
obstruction is cleared;
investigate longer runtime options of a
battery cart or alternative propulsion
system; specific EPA requirements,
cost of modification and near-term
ability to implement the choice should
be the basis for selecting the pro-
pulsion system; and
develop a "leaky coax" communi-
cation system to replace the telecom-
munications currently used. This type
of communication system utilizes the
coax cable as an antenna to transmit
and receive data.
The above mentioned enhancenru
and modifications to the prototype i
tern have been selected because t
represent the major advances neces:
to make the prototype system a fi
implementable system. These enhar
ments and modifications can be j
cessfully made with a high level
confidence.
Additionally, the prototype system '
the above mentioned enhancements
modifications should be demonstrate*
a hazardous waste site, chosen by
under the direction of the EPA, to valic
the autonomous robot's utility un
actual site conditions.
The full report was submitted
fulfillment of Contract No. 68-03-2393
Arthur D. Little, Inc., Cambridge, I
under the sponsorship of the I
Environmental Protection Agency.
Robert J. Rancatore is with Arthur D. Little, Inc., Cambridge, MA 02140-2139 and
Michael L Philips is with Advanced Resources Development Corp., Columbia,
MD 21045.
Uwe Frank is the EPA Project Officer (see below).
The complete report, entitled "Demonstration of Autonomous Air Monitoring
Through Robotics," (Order No. PB 90-134 164/AS; Cost: $17.00, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
.-c,--^ US. OFFICIAL MAIL"
4£^'1'^
/^ ' \\ •>£'•'dLTY
(O , .- , OH
I ;•• y?o U.v:,7£
_---• J /;i£S300
Official Business
Penalty for Private Use $300
EPA/600/S2-89/055
U.S.POS1AG[
? 0 2 5 =h
OOOOBSm
------- |