Vehicle
Lateral Warning, Guidance, and Control Based on Magnetic Markers
at AHSRA Smart Cruise 21 Proving Tests
Han-Shue
Tan, Benedicte Bougler
During the period
of October 6 to December 1, 2000, PATH engineers performed proving tests
on the test track of the Public Works Research Institute (PWRI) of the
Ministry of Construction in Tsukuba City, Japan. The discussions below
of the following five proving test items are related to the magnetic marker
sensing system. The depth of the discussions reflects the amount of data
available from the proving tests.
(1) Capabilities of
the PATH steering warning/guidance/control system based on the benchmark
test results to support the prevention of overshooting on curve, prevention
of lane departure under expressway speed condition, and prevention of
lane departure under fog condition.
(2) Results from the
PATH Demo 2000 scenarios to further substantiate the steering effectiveness
of the magnetic marker based lateral warning/guidance/control system.
(3) Presentation of
the proving test results of the PATH steering warning/guidance/control
system without the knowledge of the road curvature information.
(4) Comparisons of
the performance of the PATH steering warning/guidance/control system with
respect to 2m, 4m, and 6m magnetic marker spacing.
(5) Discussions on
the effects of the different magnetic markers (nail and plate types) on
the PATH and AHSRA magnetic sensing system, as well as on the noise characteristics
observed at different locations on the North Loop. The discussions in
this report are based on the preliminary observations from test results
during the short proving test period. Those results were not obtained
through an optimized process, nor do they present strong statistical significance.
Any extension of the summaries should be taken with caution. PATH tested
a vehicle that provides automatic steering control, driver lane guidance
control, and lane departure warning at the test track of PWRI in Tsukuba
City, Japan for the Smart Cruise 21 proving tests. The steering warning/guidance/control
system developed by PATH includes magnetic sensors to detect the vehicle
lateral deviation relative to the magnetic markers installed on the roadway;
an antenna to receive the radio-wave control point marker information;
an automatic steering actuator; a display screen showing the anticipated
position of the vehicle to support the driver in steering the vehicle;
and a speaker to provide an audio warning signal to the driver. This system
addresses the themes of support for prevention of overshooting on curve,
support for prevention of lane departure under expressway speed condition,
and support for prevention of lane departure under fog and rain condition.
Only the issues related to steering control are addressed; throttle and
brake are controlled by the driver. The proving tests at PWRI were performed
using one of the Buick LeSabre vehicles that were previously shown at
Demo '97, '98 and '99. The only hardware additions to the vehicle are
a human machine interface (HMI) display for lane guidance, and AHSRA's
lane marker detector for sensing system evaluation. The on-board computer
receives lateral measurements from both PATH and AHSRA magnetic sensors.
The PATH magnetic sensors were used for most tests performed at PWRI.
The AHSRA sensor data were simultaneously recorded for evaluation purpose.
The future road geometry is stored in the on-board computer and anticipated
by the communication from the AHSRA radio-wave control point markers.
The road geometry information is used, in conjunction with the lateral
deviation measurements and the vehicle dynamic model, to generate either
an automatic steering command to the steering actuator, or a display providing
a preview of the future vehicle position (predictor) if the driver does
not correct his steering action. The design of this guidance display,
which was originally developed for snowplow drivers, was optimized to
make it very easy for the driver to steer the vehicle accurately, even
in zero-visibility conditions. Furthermore, PATH also demonstrated a smooth
switching method that was previously developed, allowing the driver to
change between automatic and manual steering control at any location or
time that he commands. The lane departure warning system based on the
magnetic markers is derived from the "steering guidance" system.
A modified predictor computation triggers the departure warning signal.
A lane departure warning sounds when the modified predictor estimates
that the vehicle will reach the lane boundaries in one second. Lane departure
prevention control is a feature that the driver can select. Once activated
by the driver, the vehicle automatically switches to "automatic steering
control" if the driver continuously ignores the lane departure warning
and the vehicle begins to drift into the adjacent lane. By combining the
capabilities of automatic steering control, steering guidance display,
manual/automatic transition, as well as lane departure warning and prevention
control, the PATH lateral system based on the magnetic markers creates
a high performance lateral warning/guidance/control system. With such
capabilities, the vehicle provides a wide range of possibilities to the
driver, from guidance to full automation and from lane departure warning
to prevention control, under various curve and visibility conditions.
These capabilities also enable PATH to generate substantial test results
during the short proving test period.
The report applies
these test results to explore the five important issues related to the
magnetic sensing and control systems. The main observations from the test
results are summarized below:
(1) Automatic steering
control consistently maintains small lateral deviation (5 cm maximum on
the straight road, and 15 cm maximum on curves regardless of vehicle speeds)
as well as small curve transition error. Very small steady state lateral
deviation is observed on the curves because of the curvature information.
(2) For the steering
guidance case, the driver consistently maintains small lateral deviation
(15 cm maximum on the straight road, and 30 cm maximum on curves at various
vehicle speeds) despite the fact that the vehicle is manually controlled
with simulated zero visibility conditions. The curve transition error
is relatively small for all six curve-transitions, and so is the steady
state lateral deviation (within 10 cm) on the curves.
(3) For the case when
the driver steers based on the lane departure warning alone, the vehicle
is consistently maintained close to the lane boundaries even under severe
departure conditions. The excursions into the adjacent lane, which usually
happened on sharp curves at higher speeds, are generally kept within the
width of a tire. The error attributed to the curve transition and the
steady state lateral deviation on curves is not significant. No false
alarms or missing alarms are observed on the data sets.
A pdf of the complete
research report is available at:
http://www.path.berkeley.edu/publications/reports.htm#2001
|