Roadway Measurement System Evaluation (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

C H A P T E R 3 ResultsData Evaluation Results Following the rodeo data analysis, the research team evaluated the rodeo data elements list in relation to these four items: • Data elements reported by the participants; • Rodeo target accuracies for each data element; • Desired data element accuracies to research the causes of rural, run-off-road, single vehicle accidents as addressed in SHRP 2’s forthcoming S01E report (1); and • Best accuracy achieved by the participants during the rodeo for each data element reported. Data elements on which none of the participants reported were removed. The remaining data elements were reviewed to determine the accuracies to be sought during the S04B data collection effort. This resulted in the data elements list shown in Table 3.1. The research team also reviewed the reduced data elements list to determine those data elements necessary to answer ques- tions related to rural, run-off-road accidents. This further- reduced data elements list is shown in Table 4.1. The research team analyzed the rodeo results during the Task 7 data analysis to determine which teams could most likely provide SHRP 2 with the desired data elements to be col- lected under the S04B project. A two-tiered analysis was used covering the following: • Combined precision and accuracy of each team in report- ing each data element; and • Combined precision and accuracy of data element report- ing completeness for each team. For example, an analysis of the combined precision and accuracy for the data element “Barrier Type” results in Team 01 having a combined precision and accuracy of 164%, which results in a Team Order of 1, as shown in Table 3.2.11pleteness for each team. Team precision and accuracy for completeness are defined as follows: • Completeness precision is the average percent of the data elements included in three repetitions reported by a partic- ular team; and • Completeness accuracy is the average percent of the data elements reported in any repetition by a particular team, as compared to the number of reference data elements. Team completeness analysis resulted with Team 01 having the best completeness, followed by Team 07, Team 09, and Team 04. The teams were ranked using each of the above analysis criteria. This resulted in Team 01, Team 04, and Team 07 being at the top of the list for each criterion. Team 04 did not provide any geometric data during the rodeo; therefore, if they are to be used for data collection under the S04B project, they will need to add, and prove, this capability. Assessment of State of the Practice The current state-of-the-art in automated data collection tech- nology permits the collection of the data elements contained in the S04B Data Elements for Collection list (Table 3.1). The majority of the data elements will achieve the target accuracies contained in the original rodeo data elements list, but not all. These data elements are discussed here. Georeference of Vehicle, Images, and Roadway Features These data are readily available at the sub-1 m level of accu- racy for mobile applications. With the correct antennas and real-time differential correction signal, mobile systems can attain accuracies of sub-15 cm or sub-10 cm for the loca- tion of the vehicle and images, with slightly lower accura- cies for the roadway features extracted from the images. The georeference data for roadway features extracted fromThe analysis of the team precision and accuracy for com- pleteness in data element reporting looks at the overall com-

12Table 3.1. Data Elements for Collection in SHRP 2 Safety Project S04B Rodeo CTRE Best Target Desired Achieved Recommended Feature Data Element Definition Accuracy Accuracy Accuracy Accuracy Assets Barrier Systems On-Street parking Pavement Markings Barrier type Location Barrier beginning location Barrier ending location Barrier offset— beginning Barrier offset— ending Barrier height Post type End treatment type— beginning End treatment type—end Beginning of on-street parking Ending of on- street parking Side of street with on-street parking Location of marking—begin Location of marking—end Marking type Marking offset Centerline marking type Special pavement marking location Cable, W beam, tri-beam, box beam, concrete barrier, other Roadside or median GPS coordinates of the beginning of entire barrier system GPS coordinates of the end of entire barrier system From edge of lane to face of barrier (in.) From edge of lane to face of barrier (in.) From ground surface to top of barrier (in.) Strong post (metal [6-in. I beam]), weak post (metal [C channel, box post, 3-in I-beam]), wooden post, N/A, other Impact attenuator, buried end, terminal end, fist, bridge connection, none, other Impact attenuator, buried end, terminal end, fist, bridge connection, none, other GPS coordinates of the start of on-street parking GPS coordinates of the end of on-street parking Left, right, both GPS coordinates of start of pavement marking GPS coordinates of end of pave- ment marking Centerline; lane lines (skips); edge/fog line Offset of each type of line (cen- ter, lane and edge) from right edge of pavement. Measured to the nearest edge of mark- ing from the right edge of pavement. Broken yellow, broken/solid yel- low, double yellow, etc. GPS coordinates of edge of marking nearest to the data collection vehicle. 100% 100% ±3 ft ±3 ft ±0.25 ft ±0.25 ft ±1 in. 100% 100% 100% ±3 ft ±3 ft 100% ±3 ft ±3 ft 100% ±1 in. 100% ±3 ft 100% 100% ±3 ft ±3 ft ±0.25 ft ±0.25 ft ±1 in. 100% 100% 100% ±3 ft ±3 ft 100% ±3 ft ±3 ft 100% ±1 in. 100% ±3 ft 64% 73% ±3 ft ±3 ft -4.03 ft -2.96 ft 1.23 in. 42% 18% 15% ±3 ft ±3 ft 79% ±3 ft ±3 ft 99% 1.49 in. 100% ±3 ft 100% 100% sub-1 m sub-1 m ±3 in. ±3 in. ±1 in. 100% 100% 100% sub-1 m sub-1 m 100% sub-1 m sub-1 m 100% ±1 in. 100% sub-1 m (continued on next page)

13Table 3.1. Data Elements for Collection in SHRP 2 Safety Project S04B (continued) (continued on next page) Special pavement marking description Raised pavement markers present Location of raised pavement markers Type of roadside obstacles Offset of roadside obstacle Location of road- side obstacle Rumble strip lateral location Location of rumble strip—begin Location of rumble strip—end Rumble strip offset Location of sidewalk—begin Location of sidewalk—end Sidewalk is separated from edge of road Support type Support location Multisign Sign type(s) Location of street lighting RXR, SCHOOL, arrows, stop bar, etc. Yes/no Centerline; lane lines; edge/fog line; center and edge lines; center, lane & edge lines Tree, shrub, building, mailbox, pole, fence, stone, etc. From edge of lane to nearest point on obstacle GPS coordinates of each obstacle Centerline or shoulder GPS coordinates of start of rumble strips GPS coordinates of end of rumble strips From edge of lane to point on rumble strip nearest to the lane GPS coordinates of start of sidewalk segment GPS coordinates of end of sidewalk segment Yes/no Post, pole, sign structure, bridge, other GPS coordinates of the location where the nearest post/pole of the support enters the ground. For an overhead sign, the post/pole on the right side of the road will be used. If the overhead sign is mounted on a bridge, the location where the right-hand side of the sign is mounted. Yes/no Record the MUTCD code for each sign. If not a standard sign, record sign legend. GPS coordinates of light pole 100% 100% 100% 100% ±0.25 ft ±3 ft 100% ±3 ft ±3 ft ±1 in. ±3 ft ±3 ft 100% 100% ±3 ft 100% 100% ±3 ft 100% 100% 100% 100% ±0.25 ft ±3 ft 100% ±3 ft ±3 ft ±1 in. ±3 ft ±3 ft 100% 100% sub-2m 100% 100% ±3 ft 24% 74% 73% 39% 7.72 ft ±3 ft 100% ±3 ft ±3 ft 0.08 in. ±3 ft ±3 ft 30% 74% ±3 ft 78% 36% ±3 ft 100% 100% 100% 100% ±3 in. sub-1 m 100% sub-1 m sub-1 m ±1 in. sub-1 m sub-1 m 100% 100% sub-2m 100% 100% sub-1 m Rodeo CTRE Best Target Desired Achieved Recommended Feature Data Element Definition Accuracy Accuracy Accuracy Accuracy Assets Roadside Obstacles Rumble Strips Sidewalk Signs Street Lighting Pavement Markings (continued)

14Table 3.1. Data Elements for Collection in SHRP 2 Safety Project S04B (continued) Intersection Configuration and Dimensions Traffic Control Signalized Intersection Stop-Controlled Intersection Type of intersection Number of approaches Intersection location Number of through lanes on approach Channelization exists on approach Number of exclusive left-turn lanes Length of exclusive left-turn lane Number of exclusive right-turn lanes Length of exclusive right-turn lane Intersection has marked crosswalks Intersection is illuminated Type of traffic control Type of signalized intersection Intersection has pedestrian signal head Location of traffic signal Type of stop- controlled intersection Flashing beacon present X intersection, T intersection, Y intersection, More than 4 legs, roundabout, other 3, 4, ... approaches GPS coordinates of center of intersection. Yes/no Identify storage length of left-turn bay. If more than one left- turn lane exists, report the length of the longest one. Identify storage length of right turn bay. If more than one right-turn lane exists, report the length of the longest one. Yes/no Yes/no None, signalized, stop, yield Standard, protected left-turn, permitted turn Yes/no GPS coordinates of signal head Two-way, three-way, all-way stop control Yes/no (Flashing yellow/red beacon) 100% 100% ±3 ft 100% 100% 100% ±2 ft 100% ±2 ft 100% 100% 100% 100% 100% ±3 ft 100% 100% 100% 100% ±3 ft 100% 100% 100% ±2 ft 100% ±2 ft 100% 100% 100% 100% 100% ±3 ft 100% 100% 77% 70% sub-2m 77% 77% 67% 5 ft 67% -108.76 ft 30% 77% 77% 57% 50% N/A 77% 70% 100% 100% sub-1 m 100% 100% 100% ±2 ft 100% ±2 ft 100% 100% 100% 100% 100% sub-1 m 100% 100% Rodeo CTRE Best Target Desired Achieved Recommended Feature Data Element Definition Accuracy Accuracy Accuracy Accuracy (continued on next page)

15Table 3.1. Data Elements for Collection in SHRP 2 Safety Project S04B (continued) Bridges Approaches Driveways Lanes Median Rail Crossings Bridge begin location Bridge end location Bridge rail exists Offset of bridge rail Driveway location Driveway type Number of lanes Lane widths Location of measurement Lane add point Lane drop point Special lane function type Median type Location of measurement Median width At-grade railroad crossing location Number of tracks Railroad crossing control type GPS coordinates of where the bridge parapet or rail begins. GPS coordinates of where the bridge parapet or rail ends Yes/no From edge of lane to nearest face of rail/parapet (ft) GPS coordinates of near side of driveway Residential, farm, retail/ commercial, industrial Number of full width lanes at a location Report lane width to the nearest whole foot. GPS coordinates of reported data. Reported when the number of lanes changes, or lane width changes more than 1 foot, but not in transition reas. GPS coordinates of start of a full lane width. GPS coordinates of end of a full lane width. Two-way left-turn lane, HOV lane, bicycle lane, reversible lane, bus bay, etc. Soil, paved (striped), paved (barrier), raised curb, None, other GPS coordinates or reference post of reported data. Reported when the type changes, or the width changes more than 1 foot, but not in transition areas. GPS coordinates of first rail of first track Crossbucks, gates, flashing lights, signal ±3 ft ±3 ft 100% ±2 ft ±3 ft 100% 100% ±6 in. ±3 ft ±3 ft ±3 ft 100% 100% ±3 ft ±0.5 ft ±3 ft 100% 100% ±3 ft ±3 ft 100% ±2 ft ±3 ft 100% 100% ±0.328 ft (0.1 m) ±3 ft ±3 ft ±3 ft 100% 100% ±3 ft ±0.5 ft (0.15 m) ±3 ft 100% 100% N/A N/A N/A N/A ±3 ft 70% 33% -0.02 ft ±3 ft N/A N/A 11% 80% ±3 ft -0.13 ft N/A N/A N/A sub-1 m sub-1 m 100% ±2 ft sub-1 m 100% 100% ±0.328 ft (0.1 m) sub-1 m sub-1 m sub-1 m 100% 100% sub-1 m ±0.5 ft (0.15 m) sub-1 m 100% 100% (continued on next page) Rodeo CTRE Best Target Desired Achieved Recommended Feature Data Element Definition Accuracy Accuracy Accuracy Accuracy Roadway Inventory

16Table 3.1. Data Elements for Collection in SHRP 2 Safety Project S04B (continued) Ramps Shoulder Geometric Features Grade Cross Slope Curvature Grade of approach side of crossing Grade of leave side of crossing Ramp location Type of ramp terminal Type of section Shoulder type Shoulder paved width Shoulder total width Location of measurement Grade in direction of travel Location of measurement Location of measurement Roadway cross- slope Horizontal curve PC (point of curvature) Horizontal curve- length Horizontal curve- radius Direction (“+” uphill in direction of travel, or “-” downhill in direction of travel) and Per- cent of Slope Direction (“+” uphill in direction of travel, or “-” downhill in direction of travel) and Per- cent of Slope GPS coordinates of point of ramp gore area Entry or exit (for roadway on which the vehicle is traveling) Acceleration lane, decceleration lane, weaving section Paved, unpaved, composite (part paved, part unpaved) and curb Width of paved portion of shoul- der. Reported from edge line to edge of paved surface to the nearest foot. Total width of shoulder (composite only), including paved and unpaved parts. Measured to the first obstacle, or the break in slope. GPS coordinates of reported data. Reported when the shoulder type changes, or the width changes more than 1 foot, but not in transition areas. Direction (“+” uphill in direction of travel, or “-” downhill in direction of travel) and per- cent of slope GPS coordinates of reported data GPS coordinates of reported data Cross-slope of lane being driven. Direction (“+” slopes toward side of road or “-” slopes towards center of road) and percent of slope. GPS coordinates where curve begins ±0.5% ±0.5% ±3 ft 100% 100% 100% ±0.5 ft ±0.5 ft ±3 ft ±0.5% ±3 ft ±3 ft ±0.01% ±3 ft ±2 ft ±25 ft ±0.5% ±0.5% ±3 ft 100% 100% N/A ±0.5 ft (0.15 m) ±0.5 ft (0.15 m) ±3 ft ±0.5% ±3 ft ±3 ft ±0.10% ±3 ft ±25 ft (7.62 m) ±25 ft (7.62 m) N/A N/A > 4 m 100% 100% 100% -0.03 ft -0.29 ft ±3 ft -0.164% N/A N/A -0.2045% -154.97 ft -17.5 ft 128.48 ft ±0.5% ±0.5% sub-1 m 100% 100% 100% ±0.5 ft (0.15 m) ±0.5 ft (0.15 m) sub-1 m ±0.5% sub-1 m sub-1 m ±0.2% sub-1 m ±25 ft (7.62 m) ±25 ft (7.62 m) Rodeo CTRE Best Target Desired Achieved Recommended Feature Data Element Definition Accuracy Accuracy Accuracy Accuracy Roadway Inventory Rail Crossings (continued)

17Table 3.2. Barrier Type: Accuracy and Precision Team 01 Team 02 Team 03 Team 04 Team 05 Team 07 Team 09 Accuracy Precision Accuracy Precision Accuracy Precision Accuracy Precision Accuracy Precision Barrier Type 64% 100% 21% 78% N/A N/A 24% 89% 61% 95% 0% 92% Combined 164% 99% N/A N/A 113% 156% 92% Precision & Accuracy Team Order 1 4 N/A N/A 3 2 5digital images can be lower than the georeference of the images from which they are extracted. The difference in accuracy between image georeference and the extracted fea- ture georeference varies depending upon the placement of the feature within the digital image, as well as the care taken in calibrating the cameras on the data-collection vehicle. Roadway Assets and Features Data on roadway assets and features are readily extracted from the georeferenced digital images collected using semiauto- mated methods. Any asset that is captured in the images can be recorded and georeferenced. This includes bridges, signs, streetlights, barrier systems, trees, buildings, pavement mark- ings, and rumble strips. The quality of these data depends upon the resolution and spacing of the images, the accuracy of the GPS coordinates, the software used, the system calibration, and a rigorous QA/QC program. There have been several attempts to develop a system to perform automated image processing to extract traffic signs from digital images. These systems have had mixed results. No system is currently capable of extracting all of the roadway assets and features. Intersections The intersection attributes contained in the data elements list (Table 3.1) can be obtained from the digital right-of-way images using semiautomated data analysis methods similar to those used for recording the roadway assets and features dis- cussed. Accuracies will depend upon the GPS resolution of the mobile data-collection systems, care exercised in calibrat- ing and configuring the mobile data-collection systems, care in recording the data elements, the firm’s quality control pro- cedures, and the client’s quality assurance procedures. Geometrics The roadway geometric data elements contained in Table 3.1 are directly measured or reported by the automated data collection systems, as follows:• Grade is measured as the sliding grade representing the grade over the length of the data collection vehicle. This measure- ment can provide accuracies within ±0.5%, compared with manual measurements over the same length. • Roadway Cross-slope is typically measured between the wheel paths when using automated equipment. This equip- ment does not typically meet the 0.10% accuracy desired in the data elements list. However, using an Applanix POS LV 320 or 420 system with laser reference sensors, it is possible to achieve an average absolute error of 0.13% over 200–300 readings, compared with manual measurements, with a stan- dard deviation of 0.03%. • Horizontal Curvature data are determined from analysis of the heading data collected by the onboard inertial navigation system. The data collected from these systems, while gener- ally acceptable for network-level evaluation or for High- way Performance Monitoring System (HPMS) reporting, may not have the resolution necessary for research work. The point of curvature typically varies from the reference by ±20%, the length of curve varies by ±10%, and the radius of curvature varies by ±55%, when using all the curves that matched the reference data set. If the curve for each partici- pant that varied the most from the reference is removed from the analysis, the resulting radius of curvature varies by ±5% from the reference. While the variances may be higher than desired for research work, the systems are very repeatable, with precisions of 99%, 98%, and 91%, respectively. Many of the other data elements from the original list can also be collected, some with existing technology, some with emerging technologies, and others through new data process- ing methods. Existing technologies can be used to collect pavement roughness, which is routinely collected by state DOTs for reporting to FHWA. These same road profiling systems can be used to record pavement macrotexture. The pavement macrotexture can be measured as estimated mean texture depth (EMTD, ASTM E-965, or E-1845). If the profiling sys- tem uses 32 kHz or 64 kHz (preferred) laser profiling sensors,

18then the system can also collect EMTD. The macrotexture data is typically collected in one of the wheelpaths. Emerging technologies could be used to collect such data as pavement edge drop-off and pavement marking reflectivity, as follows. Pavement Edge Drop-off Currently there are three systems that could be used to mea- sure edge drop-off: 1. INO LRMS system that projects a line laser onto the pave- ment’s surface about 15 ft long and takes a digital image of it, which is later analyzed to determine transverse profile and rutting. If this line overlaps the edge of pavement where the drop-off occurs, it can be used to calculate edge drop-off. 2. Scanning laser system that captures a slice of the pave- ment’s surface with each scan line. These scan lines can be analyzed for edge drop-off. 3. LIDAR point cloud laser system that uses lasers to obtain a 3-D model of the roadway near the data collection vehi- cle. None of these systems has been proven for measuring edge drop-off.Pavement Marking Reflectivity The emerging LIDAR and scanning laser systems have the potential to record the varying reflectance of pavement mark- ings. Research could be performed to develop a correlation between pavement marking reflectance measured with LIDAR or a scanning laser and the pavement marking retroreflectivity measured with a pavement marking retroreflectometer, such as the Delta LTL-X. Data elements such as those dealing with horizontal and vertical curvature and sight distance can be determined using a combination of collected sensor geometric data, GPS, and images. However, an industry standard methodology to make this determination is not available currently. Reference 1. Hallmark, S., Y.-Y. Hsu, L. Boyle, A. Carriquiry, Y. Tian, and A. Mudgal. SHRP 2 Report S2-S01E-RW-1: Evaluation of Data Needs, Crash Surrogates, and Analysis Methods to Address Lane Departure Research Questions Using Naturalistic Driving Study Data. Trans- portation Research Board of the National Academies, Washington, D.C., forthcoming.