Sensor Technologies for ITS

Transcription

1 Sensor Technologies for ITS Lawrence A. Klein, Ph.D. Prepared for Transportation Research Board Freeway Operations and Signal Systems Mid-Year Committee Meeting July Park City Marriott Park City, Utah

2 Presentation Outline Sensor applications to surface street and freeway traffic management Modern traffic flow sensor technologies Theory of operation of sensor technologies Traffic parameters monitored Strengths and weaknesses of the technologies Ongoing MnDOT NIT II Sensor Evaluation Tests New sensor models Evolving vehicle detection and tracking methods Recent sensor evaluation results from TTI and Purdue

3

4 Applications of Sensors to ITS Isolated intersection control Interconnected intersection control using prestored timing plans for traffic adaptive control Interconnected intersection control using real-time data that enable traffic responsive control Freeway incident detection and congestion monitoring Traffic data collection Transponders as sources of traffic data Weather and roadway condition In-vehicle sensing, adaptive cruise control, etc.

5 Intersection Control Using Multi-Detection Zone Sensors

6 Microwave Radar Waveforms Transmitted signal Frequency Received signal Doppler frequency shift from a moving vehicle a. Constant frequency waveform Time Frequency f Transmitted Received t 1 t 2 b. FMCW Time

7 Intersection Control Using Single Detection Zone Sensors

8 Intersection Control Using VIP with Camera Mounted Over Center of Monitored Lanes

9 Component Sensors Sensors Don t Act Alone! Alternatives Inductive loop Magnetometer Microwave radar Passive infrared Laser radar Ultrasonic Acoustic Video image processing Automatic vehicle identification Automatic vehicle location Cellular telephone probes Cellular call-ins Freeway service patrol reports Call boxes/emergency telephones Closed circuit television Environmental sensors Data Processing and Display Hardware Computers Disk drives (permanent and removable) Printers Monitors Controllers Displays Video tape recorders

10 Sensors Don t Act Alone! (continued) Component Software Communications Alternatives Incident detection algorithms Ramp metering algorithms Signalized intersection control algorithms Real-time expert system (for changeable signs, ) Interface software (graphical and other types) Internal Local area network External Fiber optic Coaxial Twisted pair Microwave Spread spectrum radio Cellular telephone Citizen band radio

11 Flow rate (volume), occupancy, and density Count, presence, and passage Speed of individual vehicles and platoons of vehicles Queue lengths Traffic Parameters Approach flow profile Approach stops A given sensor generally does not output all types of traffic parameters

12 Overhead Sensor Technology Applications to Traffic Management Application Assumptions Overhead Sensor Technologies Signalized Detect stopped Microwave presence-detecting radar intersection vehicles Passive infrared control Weather not a Laser radar major factor Video image processor Signalized Detect stopped Microwave presence-detecting radar intersection vehicles control Inclement weather Signalized Detection of Microwave presence-detecting radar intersection stopped vehicles Doppler microwave radar control not required Inclement weather Real-time Desirable for Video image processor adaptive sensor footprint Microwave presence-detecting radar signal control to emulate a 6-ft Passive infrared (with suitable (e.g., SCOOT) 6-ft inductive loop aperture beamwidth)

13 Overhead Sensor Technology Applications to Traffic Management (continued) Application Assumptions Overhead Sensor Technologies Vehicle counting Detect and count Microwave presence-detecting radar (surface street vehicles traveling Doppler microwave radar or freeway) at speeds > 3 to 5 Passive infrared mi/h (4.8 to 8.0 km/h) Ultrasound Video image processor Vehicle speed Detect and count Microwave presence-detecting radar measurement vehicles traveling Doppler microwave radar at speeds > 3 to 5 Laser radar mi/h (4.8 to 8.0 km/h) Video image processor Vehicle By length Video image processor classification Laser radar Microwave presence-detecting radar Vehicle By profile Laser radar classification Inductive loop with high frequency excitation and special signal processing software

14 Sensor Technologies and Operating Principles

15 Microwave Doppler Sensors Microwave Sensors TC-20 Whelen TDN-30

16 Microwave Presence-Detecting Radar Sensors Electronic Integrated Systems RTMS Multiple Detection Zone Model Naztec 150LX Single Zone Model

17 Microwave Radar Operation Microwave Radar Antenna Power and data cables Sign bridge, overpass, pole, or mast arm mounting Path of transmitted and received energy Controller cabinet Reflected signal from vehicle can be used to determine presence (occupancy), passage (count), and speed, depending on the waveform that is transmitted by the radar sensor Vehicle

18 Microwave Radar Characteristics Parameters Count Presence (with FMCW waveform models) Occupancy (with FMCW waveform models) Speed Range (with FMCW waveform models) Instantaneous traffic density (with FMCW waveform models) Vehicle class by length (3 class limit) Installation Overhead or at side of roadway Advantages Installation and repair need not interrupt traffic Direct measurement of speed Multilane data collection Day/night operation Disadvantages With side mounting, possibility of some missed detections if tall vehicles occlude the more distant lanes (may not be significant)

19 Ultrasonic Pulse Sensors Microwave Sensors TC-30C Single Lane Model Novax ELA 155 Two-Lane Model

20 Ultrasonic Pulse Sensor Operation Ultrasonic sensors transmit and receive high frequency sound waves (25 khz to >50 khz) Vehicle range is determined using range gates, analogous to processing found in some microwave radars

21 Ultrasonic Sensor Characteristics Parameters Count Presence Occupancy Speed (with Doppler model or multiple detection zone pulse model) Range (with pulse model) Queue with multiple sensors Installation Most accurate when mounted overhead Advantages Installation and repair need not interrupt traffic Day/night operation Disadvantages Performance degraded by variations in temperature and extreme air turbulence One per lane required (one two-lane model is marketed) Low PRF may degrade occupancy measurement on freeways with moderate to high speeds

22 Passive Acoustic Array Sensors IRD SmartSonic Single Lane Model SmarTek SAS-1 Multiple Lane Model

23 Passive Acoustic Array Operation Vehicular traffic produces acoustic energy from a variety of sources such as engine noise and the interaction of the vehicle's tires with the road surface An array of microphones provides spatial directivity from which sounds are continuously detected and processed from specific locations along the roadway Signal processing algorithms confirm or reject the source of acoustic energy as a vehicle

24 Passive Acoustic Array Characteristics Parameters Count Presence Occupancy Speed (with multiple detection zones or data processing algorithm that uses an assumed vehicle length) Installation To side of monitored lanes Advantages Installation and repair need not interrupt traffic Day/night operation Multilane operation depending on model Disadvantages Cold temperature reported to affect data accuracy Specific models not recommended to detect slow moving vehicles in stop and go traffic

25 Laser Radars Schwartz Electro-Optics Autosense II EFKON TOM

26 Active Infrared (Laser Radar) Sensor Operation Detection zones are illuminated with IR energy transmitted by laser diodes Scanning beams IR energy reflected from the vehicle is focused by an optical system onto a detector array mounted at the focal plane of the optics Zone 2 Zone 1 Real-time signal processing is used to analyze the received signals and to determine count, presence, speed, and vehicle class

27 Imagery From 3-D Laser Radar Scans

28 Active Infrared Sensor Characteristics Parameters Count Presence Occupancy Speed Vehicle length Vehicle classification with 2-D and 3-D imaging models Installation Overhead (Side of roadway for toll road applications) Advantages Installation and repair need not interrupt traffic Day/night operation Multilane operation depending on model Disadvantages Performance degraded by heavy fog and blowing snow (when visibility < 20 ft)

29 Eltec 833 and 842 Passive Infrared Sensors

30

31 Passive Infrared Sensor Operation All objects emit energy based on their absolute surface temperature and emissivity at all wavelengths in the electromagnetic spectrum (Planck radiation law) Energy at infrared (IR) wavelengths can be collected by optics transmissive in the wavelength band of interest and focused on a photon detector A reference or background temperature emitted from the road surface is established The sensor's electronics detect a change in energy when a vehicle emitting an energy different from that of the background passes within the sensor's field of view. The change in energy signifies a vehicle detection. This thermal energy contrast is analogous to the visible contrast of a CCD camera in the visible spectrum

32 Passive Infrared Sensor Characteristics Parameters Count Presence Occupancy Speed (with multiple detection zone models) Queue with multiple sensors Installation To side of monitored lanes Advantages Installation and repair need not interrupt traffic Day/night operation Potentially better performance with long wavelength IR than visible wavelength sensors in some fog conditions Disadvantages Performance degraded by heavy rain, fog, snow One per lane required

33 Video Image Processors: Traficon VIP-2

34 Autoscope 2004 and Peek Video Trak-900

35 Iteris Vantage VIP Family

36 Video Image Processor (VIP) Operation Video image processor field of view determined by focal length of lens and camera mounting height. Some VIPs insert vehicle detection zones into the camera s field of view based on traffic management and data collection requirements; others track vehicles through the entire field of view.

37 VIP Characteristics in Upstream and Downstream Viewing Upstream Viewing Headlight blooming, glare on wet pavement, headlight beams detected in adjacent lanes on curved road sections More blockage from tall trucks Traffic incidents not blocked by resulting traffic queues Downstream Viewing Cameras on overpasses concealed from drivers More information from tail lights available for braking indication, vehicle classification, turning movement identification, and tracking Easier to acquire vehicles that are closer to the camera for tracking algorithm implementation

38 Video Image Processor Characteristics Parameters Count Vehicle and queue length Presence Vehicle classification by length (up to 3) Occupancy Alarms Speed Expanded traffic parameter data set Installation Overhead or to side of roadway Advantages Installation and repair need not interrupt traffic Single camera and processor can service multiple lanes Rich array of data available Disadvantages Large vehicles project their image into adjacent lanes, sometimes leading to false detection; large vehicles can also mask trailing vehicles Shadows, reflections from wet pavement, day/night transitions, headlight beams, relative color of vehicles and background, camera vibration can affect detection Cost 50 to 70 ft camera height for optimal side viewing performance

39 ASIM Technologies Sensor Combinations DT 281 Infrared-Doppler radar sensor DT 272 Infrared-ultrasonic sensor

40 Midian Electronics SPVD-2 Magnetometer Sensor SPVD 2 Sensor/ Transmitter Type Channel Receiver NEMA TS-1 Style 1-4 Channel Receiver One-Channel Receiver

41 Nu-Metrics Groundhog Magnetometer Sensors G-1 G-2 and G-2wx

42 Safetran Magnetic Sensor 231E Sensor Probe 232E Sensor Electronics

43 3M Microloop Probes (Passive Magnetic Sensors) Model 701 Model 702

44 Magnetometer Operation cl Subsurface magnetometers Shoulder Magnetometers consist of one or more turns of wire wound around a magnetic core material Magnetometers sense the presence of a ferrous metal object by the perturbation it causes in the Earth's quiescent magnetic field Magnetometer array Magnetometers can be used on bridge decks where ILDs may be affected by the steel support structure or simply cannot be installed Three-axis fluxgate magnetometers can be arrayed to give vehicle signatures in support of vehicle classification

45 Magnetometer Characteristics Parameters Count Presence Occupancy Average vehicle speed (with two sensors in a speed trap configuration) Queue length (with multiple sensors) Classification (with sensor arrays and special signal processing) Installation Embedded in roadway Advantages Low per unit cost Can detect small vehicles including bicycles Arrays of magnetometers can provide vehicle classification Disadvantages Traffic interrupted for installation and repair (exceptions exist) Difficulty in discriminating longitudinal separation between closely spaced vehicles

46 Inductive Loop Detector Vehicle Classifiers Reno A&E S-1500 Series Inductive Loop Vehicle Classifier and Speed Sensor Peek Traffic Axle Location and Vehicle Classification System

47 Traffic Sensor Output Data, Bandwidth, and Cost Technology Inductive loop Magnetometer (Two-axis fluxgate) Magnetic (Induction coil) Microwave radar Active infrared Passive infrared Ultrasonic Acoustic array Video image processor Output Data Multiple Lane, Multiple Detection Zone Data X X X 2 X X 3 Low to moderate X X X 2 X Low Count Presence Speed Occupancy Classification Communication Sensor Purchase Cost 1 Bandwidth (each in 1999 $) Low 9 ($500 to $800) Moderate 9 ($900 to $6,300) X X 4 X 2 X Low Low to moderate ($385 to $2,000) X X 5 X X 5 X 5 X 5 Moderate Low to moderate ($700 to $3,300) X X X 6 X X X Low to Moderate to high moderate ($6,500 to $14,000) X X X 6 X X X X Low to moderate Low to moderate ($700 to $1,200) Low X X X X X 7 Low to moderate X X X X X X Low to high 8 Low to moderate (Pulse model: $600 to $1,900) Moderate ($3,100 to 8,100) Moderate to high ($5,000 to $26,000) 1. Installation, maintenance, and repair costs must also be included to arrive at the true cost of a sensor solution. 2. Speed can be measured by using two sensors a known distance apart or by knowing or assuming the length of the detection zone and the vehicle. 3. With specialized electronics unit containing embedded firmware that classifies vehicles. 4. With special sensor layouts and signal processing software. 5. From microwave radar sensors that transmit the proper waveform and have appropriate signal processing. 6. With multi-detection zone passive or active mode infrared sensors. 7. With models that contain appropriate beamforming and signal processing. 8. Depends on whether higher-bandwidth raw data, lower-bandwidth processed data, or video imagery is transmitted to the traffic management center. 9. Includes underground sensor and local receiver electronics. Electronics options are available for multiple sensor, multiple lane coverage. 9

48 Advantages and Disadvantages of Traffic Flow Sensor Technologies Technology Advantages Disadvantages Microwave Doppler Microwave True Presence Good performance in inclement weather Direct measurement of speed Good performance in inclement weather Detects stopped vehicles Can operate in side-looking mode to service multiple lanes Passive Infrared Provides day and night operation Multizone passive sensors measure speed Active Infrared Direct measurement of speed Provides vehicle classification data Cannot detect stopped or very slow-moving vehicles Performance possibly degraded by heavy rain, fog, overcast skies, or snow Performance degradation by heavy fog [visibility < 20 ft (6 m)] and blowing snow Installation and maintenance require lane closure Ultrasonic Compact size, ease of installation Performance may be degraded by variations in temperature and extreme air turbulence Low PRF may degrade occupancy measurement on freeways with mod to high speeds Visible VIP Single camera and processor can service multiple lanes and multiple zones/lane Rich array of traffic data available Easy to add and modify detection zones Some vehicle occlusion may occur with side-looking, multiple lane sensor Large vehicles can mask smaller vehicles, leading to undercounting Tall vehicles can project their image into adjacent lanes, leading to overcounting Shadows, reflections from wet pavement, vehicle/road contrast, headlight projection into adjacent lanes on curved road sections, day/night transitions, camera vibration, and debris on camera lens can affect performance Side viewing requires high, stable camera mounting platform Over-roadway camera mounting requires lane closure for installation and maintenance

49 Infrared VIP Advantages and Disadvantages of Traffic Flow Sensor Technologies (continued) Technology Advantages Disadvantages Possibility of using same algorithms for day and night operation Rich array of traffic data available Low-cost technology not yet available, but being developed Acoustic Insensitive to precipitation Cold temperature has been reported as One model services multiple lanes affecting data accuracy Specific models not recommended to detect slow moving vehicles in stop-and-go traffic Magnetometer Magnetic (passive) Inductive Loop Detector Less susceptible than loops to stresses of traffic Some models transmit data over wireless RF link Can be used where loops are not feasible (e.g., bridge decks) Some models installed under roadway without need for pavement cuts Less susceptible than loops to stresses of traffic Standardization of loop electronics units Excellent counting accuracy Mature, well understood technology Some models provide classification data Installation requires pavement cut Installation and maintenance require lane closure Decreases pavement life Small detection zone Installation requires pavement cut or tunneling under roadway Cannot detect stopped vehicles (exception for 1 model using multiple sensors and application specific software from vendor) Reliability and useful life are dependent on installation procedures Installation and maintenance require lane closure Decreases life of pavement Susceptible to damage by heavy vehicles, road repair, and utilities Multiple detectors usually required at a site

50 Overhead Sensor Data Compatibility In Existing Systems Frequently asked questions concerning sensor compatibility are: Are sensor connections compatible with NEMA TS1/TS2 and Model 170/2070 controllers? Does the sensor have a loop detector emulation mode (contact closure)? How should the sensor cable be terminated in the cabinet and how should it be connected to the controller? Answer: Most overhead sensors provide outputs compatible with inductive loop detector outputs. Manufacturer s specifications usually include the interfaces and communications protocols they support. Specific inquiries can be made of the sensor manufacturers. Serial data interfaces require software drivers to be written.

51 Processing of Sensor Data Sensor selection and data processing are dependent on the application and algorithm selected to support the application Incident detection (e.g., California, McMaster, speed variance) Arterial signal control (e.g., UTCS, CIC, SCOOT, SCATS) Ramp metering (e.g., Local Responsive such as ALINEA, Wide-Area Responsive)

52 Evolving Vehicle Detection and Tracking Methods Tracking cellular telephone callers while they are in vehicles Location and speed measurement accuracies (U.S. Wireless): In a test of vehicles using AMPS and IS-136 handsets, location accuracy was within a median location of between 40 and 50 meters (60 to 70 meters using the FCC standard of measuring accuracy at the 66 percent level). Speed accuracies were within 2 to 4 percent of actual speeds for speeds below 75 miles/hour. Sensor data acquisition from aircraft, satellites, and UAVs Pitu Mirchandani of the Univ. of Arizona and National Consortia on Remote Sensing in Transportation

53 Results From Some Current Sensor Evaluation Studies Dan Middleton of TTI: Reports on evaluation of RTMS, Iteris Vantage, Autoscope Solo Pro, and SAS-1 at a freeway testbed in Austin Darcy Bullock of Purdue University: Reports on VIP performance at unlighted signalized intersections in Indiana