Moving Vehicle Characterization With Ultra Wideband (UWB) Radar
CASIS 2004Peter HaugenUCRL-PRES-208064
Acknowledgements:Co-author:Richard LeachLLNL MIR groupGreg DallumSteve FreemanCarlos RomeroPat WelshSpecial Thanks:Union Pacific Railroad
Application Scenarios
Potential sponsors at DOD (Army, Navy) and DHS have interest in automated detection, characterization, and assessment of vehicle traffic. (trains, trucks, tanks, autos, etc.)
Agend
a
1. Project Goals and Restrictions2. Sensor Selection and Descriptions3. Sensor Testing and Performance4. Data Collection5. Data Analysis
Topic Areas:
Project Goa
ls
Sensor Readings
Signal Processing
Presence of a target (train or automobile)
Direction of target travelSpeed of target
Location of target (lane/track of travel)
Target car count (train)
Vehicle/train car identification
Project Restrictions
• Low power, battery operated• Rapid set-up• Remote data processing and/or exfiltration• Operate in a variety of weather conditions and
times of day• Standoff from road/tracks• Low cost, disposable
Agend
a
1. Project Goals and Restrictions2. Sensor Selection and Descriptions3. Sensor Testing and Performance4. Data Collection5. Data Analysis
Topic Areas:
UWB is the optim
al choice
Medium
Low
High
Low
Cost
MediumClearReasonable, insufficient
DifficultOptical beam
LongAllReasonableDifficult, no standoff
Magnetometer/ Geophone
ShortClear, Day
DifficultDifficultCamera
LongAllReasonableEasyUWB Radar
LifetimeWeatherData Complexity
Opera-tion
Red - UnacceptableYellow- Meets all requirements
Sensor Options and Comparison
Data C
ollection, Fixed-Range
• Fixed-range sensors look for motion at a specific distance from the sensor
• Two are combined to measure target speed and direction
Fixed Range Radars Location
Fixed-Range Radars
Fixed-Range A
nalysis
Fixed-Range radar time series provides: Velocity
• Low power target detection
• Velocity measurements
• Potential car count
• Some discrimination information
Time
Da
ta C
ollection, Swept-Ra
nge
• Swept-range radar act like a type of range finder, giving radar reflectivity as a function of distance
• Profiles the side of the targetSwept-Range Radar
Swept-Range A
nalysis
Swept-range radar estimates:
• Detection
• Car count
• Discrimination
• Location
• Speed *
• Direction *
* using 2 radar
Time
Agend
a
1. Project Goals and Restrictions2. Sensor Selection and Descriptions3. Sensor Testing and Performance4. Data Collection5. Data Analysis
Topic Areas:
UWB Sensor Testing
48cm 30cm 15cm 8cm
150cm100cm
150cm
Diagram of experimental setupDiagram of experimental setup
LINEAR SLIDE
TARGETS 4x4cmUnistrut material (3)CONCRETE SLAB FLOOR
RADAR UNIT
UWB Sensor Testing
UWB Radar performance evaluation and verification test
Radar tested by scanning a series of metal targets while the heights and distance between the targets is increased
Swept Range Results
Analysis of the swept-range sensor data
100
50
0
100
50
0
100
50
0
100
50
00 0.5 1.0 1.5
distance (m)
3 lengths of 4cm square stock centered at 8cm spacing (red)
3 lengths of 4cm square stock centered at 15cm spacing (red)
3 lengths of 4cm square stock centered at 30cm spacing (red)
3 lengths of 4cm square stock centered at 48cm spacing (red)
corresponding radar images
600
400
200
0
-200
-400
-6000 0.5 1.0 1.5
distance (m)
Inte
nsity
Agend
a
1. Project Goals and Restrictions2. Sensor Selection and Descriptions3. Sensor Testing and Performance4. Data Collection5. Data Analysis
Topic Areas:
Deploym
ent Setup
Cameras and Speed Radars were used in data collection to validate the sensor measurements and provide ground-truth for the signal processing
Deploym
ent Data Results
UWB Radar ReturnsSignal Processing
Desired Statistics
Time, radar frames, train carsStart of train
End of train
Distance from
Rad
ar 63 Rail Cars
Agend
a
1. Project Goals and Restrictions2. Sensor Selection and Descriptions3. Sensor Testing and Performance4. Data Collection5. Data Analysis
Topic Areas:
Data A
nalysis
UWB Radar Returns
Signal Processing
Presence of a target (train or automobile)
Direction of target travelSpeed of target
Location of target (lane/track of travel)
Target car count (train)
Vehicle/train car identification
Data A
nalysis
Further data analysis details available. Refer to the LLNL Micro-Power Impulse Radar group for additional information.
Summ
ary
Our research has shown that a low power, UWB radar-based hardware/software combination can be used to autonomously detect, characterize, and identify target trains and potentially other types of vehicles.
=
Questions?