Weather Applications and Products Enabled Through Vehicle Infrastructure Integration (VII)
5. Weather-Related Vehicle Data
PROSPECTIVE DATA ELEMENTS: USES AND ISSUES
The automotive industry is making significant technological advancements in the areas of vehicle environmental sensing and vehicle responsiveness to road conditions. Because of these developments, direct measurements of environmental variables such as temperature and pressure are becoming routine. Other variables presently available on vehicles such as wiper setting, anti-lock brake status, and stability control status, have the potential to address weather-related safety and mobility challenges that motorists experience on a daily basis. It is also expected that continued innovation within the automotive sector will provide opportunities to measure or derive additional atmospheric and road condition parameters.
The FHWA Road Weather Management Program identified and published a comprehensive list of road weather related variables that have a considerable impact on roads, traffic flow, and operations (1). Improvements in weather and road condition observations and forecasts are needed to minimize the impact of weather on the roadway system. Can vehicle data contribute to the solution?
The first column in Table 5.1 contains a list of the high impact road weather variables noted by the FHWA Road Weather Management Program. The second column lists the corresponding vehicle data elements that have been identified as having the potential to contribute to the diagnosis and prediction of each road weather variable. The final two columns highlight the challenges and issues associated with using the vehicle data and provide additional commentary regarding selected elements. The information supplied in the table assumes that vehicle location (GPS latitude and longitude) and time will be transmitted along with the vehicle data elements noted in the table.
The challenges and issues presented in Table 5.1 highlight the need to account for possible differences in the sensors types used by automobile manufacturers, along with variations related to where the sensors are placed and their primary function, as these factors will influence data quality and accuracy. It may also be necessary to develop a comprehensive understanding regarding the way in which vehicle operators interact with and use on-board systems. This information could be used to refine products that are based on vehicle data. Finally, not all vehicle data will be available in the quantities needed to support the development of new applications or impact existing applications; it will be several years before certain variables (e.g. pavement temperature) are widely available on vehicles. These challenges and issues are discussed in further detail in Section 10 of this document.
The following sections provide a more comprehensive look at some of the vehicle data elements presented in Table 5.1.
Vehicle measurements of outside air temperature are commonplace as this data element is required for the efficient operation of the vehicle emission system. Many vehicle models have multiple temperature sensors. In fact, in many of today's production vehicles, the operator is presented with a digital readout of the current ambient temperature. In some cases, the temperature presented to the operator is from a different sensor than that used for engine performance. Ambient temperatures derived from mobile platforms could be used to enhance weather-related products for the surface transportation industry. However, utilizing these data will not be without its challenges and caveats.
Automobile manufacturers acquire temperature sensors from a variety of original equipment manufacturers (OEMs); therefore, temperature readings from two separate vehicles within close proximity of each other may differ slightly because of the sensing characteristics of the devices. This may also hold true for vehicles of the same make. Moreover, without some type of standard in place for VII probe data, temperature measurements that come from vehicles could be taken from different physical locations on each vehicle. For instance, one automobile manufacturer may provide temperature values associated with the air intake sensor on the vehicle, while another could supply measurements from a sensor mounted behind the radiator grille and in front of the engine compartment, as is the case with the Jeep Grand Cherokee. Some vehicle models (e.g., Honda Pilot) have a temperature sensor in the front bumper. A recent investigation by Mitretek on vehicle-based air temperature measurement accuracy showed that sensor placement has a significant impact on air temperature measurements (13).
Sensor responsiveness to environmental temperature change is also an important factor. An ambient air temperature sensor on a vehicle that has been housed in a garage will take some period to adjust to the outside air temperature once the vehicle exits the garage. The larger the initial temperature difference, the longer the response period. Temperature values displayed to the operator may also be filtered by internal software in an effort to dampen out temperature oscillations and provide a level of continuity more in line with the vehicle operator's expectations. If these values are collected by the vehicle and transmitted to the VII network, some spatial variations in the surface temperature field will be lost.
It has been shown that temperatures retrieved from mobile sensing platforms can also be impacted by both idle time and heavy traffic (13). The measured ambient air temperature can exhibit a warm bias under these conditions; therefore, knowing how long the vehicle has been in operation and the traffic conditions in which the vehicle is operating at the time the measurement is taken will be essential to making effective use of these data.
Knowing the relative humidity at scales that VII would enable will have considerable implications on the diagnosis and prediction of several high impact weather and road conditions. There is little doubt that relative humidity data in sufficient quantities would almost instantly result in improvements in the analysis and forecast of precipitation and fog. However, this variable is not currently available on the majority of production vehicles. Notable achievements in the automotive industry regarding the measurement of relative humidity are related to measurements taken on the inside of the front windshield. Data from humidity sensors on the inside of a windshield are being used to anticipate the formation of moisture on the windshield; as a result, the climate control system can be automatically activated and adjusted to mitigate windshield fogging. Humidity measurements of this type will not directly benefit the development of weather-related applications, but advancements in on-board cabin relative humidity measurements may lead to the production of low cost exterior devices that assess and monitor atmospheric moisture content in support of automotive applications such as engine efficiency applications.
Knowledge about the water vapor content of the atmosphere, whether it is derived from dew point or relative humidity measurements, is critical, as water vapor is an atmospheric state variable that directly influences the formation of clouds, precipitation, and fog. Water vapor is one of the most variable measurements in space and time. Slight changes in its value over small temporal or spatial scales can have a large impact on weather and road conditions. Improving the measurement of boundary layer moisture is a very high priority of the atmospheric science community. Because the atmosphere is very sensitive to water vapor, sensing systems must be very accurate. This poses a challenge to the automotive industry because accurate water vapor sensors are more expensive than temperature sensors. The benefit of having accurate water vapor measurements is high, so an investment in research in this area would be worthwhile.
Because the price of water vapor sensors is high, it may not be practical in the short term to expect vehicle manufacturers to instrument vehicles with these sensors. Another possible approach would be to instrument a subset of vehicles with water vapor sensors. Vehicle fleets that have broad coverage on a daily basis would be good candidates for these sensors. Fleets that have broad coverage on a daily basis could include vehicle fleets that deliver mail or packages, for example. State maintenance or emergency vehicles may also be candidates for additional atmospheric sensors.
Wiper state would be another extensively used variable because it is widely available, and can provide information about the state of the atmosphere (precipitation) and roadway (wet or dry).
A key factor in using information linked to wiper state is understanding how different vehicle operators interface with and use the wiper settings under disparate conditions. Research conducted by the University of Michigan suggests that middle-aged drivers used the wiper system the most, while older drivers reduced their speed and increased their headway time margin (time to the preceding vehicle) once the wipers are on (14). Further research is necessary in order to fully document how wipers are used during adverse weather conditions. Due to privacy concerns, specific information regarding the driver will not be available through VII, but it may be essential for application and product developers to account for variations in the way populations use windshield wipers.
From an application development standpoint, wiper state will be useful in determining where precipitation is occurring, as it would provide a binary indication (yes or no) of the presence of rain or snow. However, it is likely that wiper state data will need to be used in conjunction with ancillary data to identify areas of precipitation. Windshield wipers are not only used during precipitating conditions, but they are employed when roadways are wet and no precipitation is occurring. Driving in moderate to heavy traffic conditions on wet roadways sometimes requires the use of low or intermittent wiper settings because of the existence of roadway spray. These conditions generally occur after the passage of rain/snow or in between showers. Operators also use wipers when washing the windshield. If used improperly, wiper state could lead to a misdiagnosis of precipitation occurrence. Because most automobile crashes occur under wet pavement conditions, the characterization of roadways (wet or dry) using wiper state data will also be valuable, but it too will likely require supplemental data.
Some of the difficulties associated with the use of wiper state data would be alleviated to some extent by the proliferation of rain sensors. Rain sensors generally operate on the principle that the accumulation of water on the outside of a windshield will disrupt an infrared beam of light that is emitted by the sensor, which is mounted on the inside of the windshield. The magnitude of this disruption can be correlated to the amount of water on the windshield. Information from these sensors is used to control the operation of windshield wipers, including the speed and interval of the wipers. Rain sensors would provide accurate, objective data related to wiper usage. Nonetheless, the sensors will not solve the problem of discriminating between precipitation and road spray; the sensor functions solely on the basis of water on the windshield, and does not account for the source of the water.
It is also not clear how well rain sensors function when it is snowing or if they will falsely report rain when ice and snow melt off the windshield. Additionally, rain sensors may falsely indicate the absence of precipitation when vehicles drive under bridges or enter tunnels, even though heavy precipitation may be occurring. In these situations, the amount of water on a vehicle's windshield would be considerably reduced, and data from the sensor would suggest a lack of precipitation at that location. Furthermore, these data would turn off the wipers or significantly decrease wiper speed.
5.6 Lights (Fog and Headlights4)
As with windshield wipers, information associated with vehicles' outside lighting systems would be broadly accessible through VII. It is anticipated that information regarding light status would be used to infer conditions related to the current state of the atmosphere (e.g. rain, snow, fog, darkness, etc.). Again, the human factors aspect is significant with respect to how drivers use headlights and fog lights. Although all drivers are taught to turn on their lights when precipitation is occurring, visibility drops significantly, or near sunset, not all drivers adhere to these standards. Even drivers that do their best to follow these guidelines do not operate headlights or fog lights in a consistent manner. Increasingly, vehicles are equipped with devices (e.g. sun and rain sensors) that are capable of operating the lights without the assistance of the driver which would minimize the subjective nature of headlight and fog light operation. Utilizing headlights or fog lights to derive information about the environment will require a rich set of vehicle observations as well as ancillary data (weather radar, satellite, surface observations, etc.).
Vehicle acceleration, braking and turning results in forces that can be sensed and measured by an accelerometer. One of the most notable uses of accelerometer data is to deploy air bags when a vehicle undergoes a rapid deceleration. These data can also be used to initiate other automotive systems such as stability control. A three-axis accelerometer can provide information regarding forces along the longitudinal, lateral and vertical axes. In terms of basing weather applications and products on these data, it will be necessary to use accelerometer data in conjunction with other vehicle data elements. Additionally, a significant amount of research and development will be required to make effective use of these data, and product developers will need to work closely with OEMs to gain an adequate understanding of the issues related to accelerometer data.
As indicated in Table 5.1, it may be possible that data from accelerometers could be used to derive information about wind and road conditions. During high wind events, accelerometer data, along with data associated with steering wheel inputs and nearby surface meteorological stations, could be used to infer areas of high winds blowing perpendicular to a roadway. Many drivers have experienced conditions where their vehicle was subjected to high winds, and as a result, sudden vehicle accelerations (longitudinal and/or lateral) occur, which require abrupt steering inputs from the driver or moderate steering force to maintain control and the desired heading. It is very unlikely that currently available vehicle data can be used to accurately derive wind speed and direction; however, providing high wind alerts to drivers, especially operators of high profile vehicles, would be extremely valuable.
The discussion contained in the following section supplies additional background on how accelerometer data will benefit VII-enabled analyses.
The Anti-lock Braking System, or ABS, was introduced on production vehicles approximately 20 years ago. During deceleration, ABS prevents the wheels of a vehicle from locking up in an effort to maximize traction. The advent of ABS was followed by traction control, which also works to maximize traction during acceleration by minimizing wheel spin. However, traction control systems only function at low vehicle operating speeds. In recent years, a growing number of production vehicles have come with stability control. Stability control uses a combination of throttle and braking action to control the lateral movement (yaw) of the front and/or back of a vehicle in an attempt to keep the vehicle from getting into an uncontrolled skid.
Each of these systems is designed to monitor wheel events linked to the pavement/tire interface. By making appropriate adjustments to other onboard systems (e.g. throttle, braking, etc) the greatest amount of traction possible is achieved. Wheel lock, spin, and sideslip by and large are more likely to occur during adverse road conditions when surface friction is low. Thus, information gathered through VII concerning the actuation of these systems would be able to aid in diagnosing slippery road conditions (e.g. wet, snow covered, icy).
To get an accurate assessment of road conditions using ABS, traction control, and stability control data, it will be necessary to gather supplementary data related to the vehicle's motion. It is possible to record an ABS event when roadways are dry. For instance, an ABS event may occur on a dry roadway when a driver is required to come to an abrupt stop to avoid an accident. During this type of scenario, a significant amount of brake pressure (100% brake boost) is usually applied and the deceleration recorded by an accelerometer is relatively high. In a situation where an intersection is snow-covered and a driver is required to stop for a red signal light, an ABS event is less likely to be accompanied by strong braking and a sharp deceleration. Of course, a vehicle stopping on a dirt or gravel road may generate similar sensor reports; therefore, information on the road surface may be required to differentiate poor road conditions.
It has been long recognized that solar radiation (insolation) is a critical component in forecasting road temperature. Automotive innovation has led to the increased use of sun sensors to automatically control the heating, venting, and air condition systems on vehicles by detecting the strength of sunlight entering the cabin of the vehicle. Data obtained from these sensors through VII will lead to a more accurate assessment of sky conditions, which would translate into more accurate road temperature forecasts. Presently, sun sensors are available on higher end automobiles, so it may be some years before there is enough data available from these sensors to improve road temperature predictions.
A considerable amount of research is being conducted to develop systems that will assist vehicle operators in the identification and avoidance of driving hazards. These systems include the use of radar and video technology. An example of this can be seen in the emergence of Adaptive Cruise Control (ACC) systems, which utilize millimeter-wave radar (76-77 GHz) to detect vehicles in front of the host vehicle and adjust the host vehicle's speed accordingly to maintain a safe driving gap. Although radar technology used in driver assist systems is not largely impacted by adverse weather by design, it may be possible to use these data to detect some weather-related driving hazards (e.g. fog).
Millimeter-wave (MMW) radar systems are used in a wide range of applications today including adaptive cruise control, automobile collision warning, missile guidance, speed measurement, and clear air turbulence. The characteristics of the radar vary greatly depending on the use. More expensive millimeter-wave radars are pulsed and may include a Doppler capability. Less expensive radars utilize continuous wave (CW) technology including frequency modulated continuous wave technology, which has an excellent ability to measure target range.
The portion of the electromagnetic spectrum that comprises the MMW region ranges from 30 to 300 GHz or wavelengths of 10 mm to 1 mm (15). Atmospheric propagation effects play a large role in the design and utilization of radar applications operating in the MMW region. Propogation effects include absorption, attenuation, backscatter, phase variation, polarization, ducting, arrival angle variations, and surface phenomena (16).
The attenuation5 of MMW at sea level and 4 km altitude is displayed in Figure 5.1. There are several frequencies that exhibit minimal atmospheric attenuation (35, 95, 140, and 220 GHz); however, other frequencies are significantly impacted by oxygen and water vapor. Figure 5.2 displays the impact of rain and fog on the propagation of MMW energy. These data indicate that losses due to rain are greater than fog. This characteristic is supported by a study conducted by the University of California. The investigation found that precipitation in the form of rain and wet snow had the largest attenuation effects when compared to dry snow and fog, which had much smaller effects (16). Although driver assist systems that use MMW radar have been designed to mitigate the impact of adverse weather, it is possible that MMW data could contribute to the diagnosis of hazardous weather, especially at the 24 GHz frequency, which has been targeted for short-range driver assist applications such as blind spot detection applications, which aids in the detection of objects in vital zones around the vehicle.
Video data would act as a surrogate for radar data in the detection of adverse weather. As advancements associated with driver assist video technology are realized, timely and accurate assessment of road weather hazards will be made possible through pattern recognition techniques.
Vehicles are currently capable of contributing data that would lead to improvements in road weather applications and products. Presently, weather-related vehicle data elements range from direct measurements of environmental conditions (e.g. temperature) to indirect indications of road conditions (e.g. ABS). Further advancements in the automotive industry will likely result in additional data that could help facilitate subsequent advances in the timeliness and accuracy of road weather products. Due to issues such as variations in sensor type, sensor placement, and principal function, there will be a need to develop a comprehensive understanding of each weather-related vehicle data element. This understanding would ensure appropriate and effective use of VII-enabled data.
4 Assumes that taillights are active when headlight switch is in the on position. In this document, running lights are not considered, as they are not likely to support road weather applications.