Use of Automated Machine Guidance within the Transportation Industry (2018)

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.

NCHRP Project 10-77 95 CHAPTER 9: ACCURACY OF AMG PROCESSES During various stages of the project life cycle, accuracy is a key issue, particularly surrounding the following areas of the project: • Initial data collection for developing existing surface terrain • Development of DTM and EED • AMG processes, procedures, and end-user competencies • QA/QC reported practices • Heavy and fine grading equipment operations • Paving equipment operations The survey conducted as part of this research specifically targeted these areas to gather input from contractors, software/hardware vendors, and agencies (See Chapter 3). Accuracy issues involving initial data collection and DTM/EED modeling are discussed in Chapter 6 and Chapter 7. This chapter focuses on the several factors that influence the accuracy of the AMG process. As background information, findings of a discussion session from the 2009 workshop on AMG accuracy (with cross-references to the outcomes from the survey) are presented below. (See Appendix A for full workshop report.). ACCURACY REVIEW – AMG WORKSHOP A session at the AMG workshop was dedicated to the discussion of AMG accuracy issues. Participants were asked to recall instances where AMG accuracy was compromised. They provided feedback on various sources of errors that contributed to the overall accuracy of AMG, as summarized in Table 9-1. Table 9-1. Various Sources of Errors Contributing to the Overall AMG Accuracy (2009 Workshop Findings) Error # Description (random order) 1 Errors in setting up the control network 2 Inaccuracy in the preconstruction survey used to develop the DTMs 3 Errors in the design 4 Errors that result from faulty software 5 Errors that result from faulty hardware 6 Limitations to the accuracy of positioning method (GPS, total station, or laser) 7 Errors transmitting control information from the positioning equipment (GPS, total station receiver, or laser) to the machine hydraulic controls for the ground- or pavement-engaging equipment 8 Inability of machine hydraulic controls to respond accurately or smoothly to instructions from AMG units (overcorrection, slow response, and other similar problems) 9 Human error in operating hardware, software, and equipment 10 Failure to identify inaccuracies during the QA/QC process (or false indications of inaccuracy during a QA/QC process) 11 Sensor/technology/system limitations (such as pushing beyond the limits of the equipment)

NCHRP Project 10-77 96 Frequency of the errors (by Error # shown in Table 9-1) and strategies for error detection and mitigation are detailed in Table 9-2. The research team also conducted interviews with various contractors after the workshop to get feedback on the various error detection and mitigation strategies. Some contractors shared their field experiences on various sources of errors, and these are summarized in the Table 9-2 notes. Survey outcomes from contractors, vendors, and agency personnel on factors contributing to the overall accuracy of AMG are also cross-referenced in the Table 9-2 notes. VARIABLES THAT INFLUENCE THE ACCURACY OF AMG PROCESSES The accuracy of the AMG process is primarily influenced by three variables: • Position measurement technology • Construction process • Human errors Survey responses from surveyors and planners indicated total station surveying (robotic and conventional) is considered more accurate than GPS and photogrammetric surveying. Manufacturers and researchers have published the precision and accuracy values of various position measurement technologies in the technical literature (Peyret et al., 2000; Retsher, 2002; Barnes et al., 2003; Mautz, 2008; and Trimble, 2008). It does not appear that the effect of construction process and human errors has ever been thoroughly studied or quantified. Most contractors, vendors, and agency personnel who responded to the survey questions reported that these variables play a significant role in the overall accuracy of the AMG process (See Table 9-2). The remainder of this chapter summarizes some of the key factors that affect these variables and a statistical approach along with experimental results to assess factors affecting accuracy. Position Measurement Technologies Table 9-3 provides a summary of accuracy, coverage range, measurement principle, and relative cost of different position measurement technologies that are typically used in construction applications. The laser or ultrasonic technologies offer higher vertical (elevation) accuracies than GPS and have shown success in achieving tighter tolerances on some fine grading projects (Daoud, 1999). However, laser or ultrasonic technologies have some practical limitations with use in rain, dust, wind, and snow, and need frequent charging of deep cell batteries (Cable et al., 2009). These technologies also require a direct line of sight between the control station and the receiver on the equipment, which is why they have not been used on heavy earth moving equipment, other than motor graders (Jonasson et al., 2000). GPS-based technologies can overcome the limitations stated above with laser and ultrasonic technologies, but they don’t offer high vertical accuracy. Based on field studies, Peyret et al. (2000) noted that RTK GPS systems normally have vertical accuracy (±20 mm) or twice the horizontal accuracy (±10 mm). A vertical accuracy level of ±20 mm is not sufficient for applications such as paving. Peyret et al. (2000) proposed a solution to improve the RTK GPS vertical accuracy to ±10 mm by filtering (or post- processing) high-frequency noise and low-frequency bias in the GPS signal. The filtering approach, however, posed some problems in the field, especially when conditions (such as constellation, horizon, multi-paths, or masking) are different from one antenna to the other.

NCHRP Project 10-77 97 Table 9-2. Frequency of Errors and Suggested Detection/Mitigation Strategies Error # Frequency Detection Mitigation Strategy Daily Seldom Project Duration Random 1 √ F Field QC Training and standardized protocols 2 √ Design phase Use of advanced surveying techniques, such as digital photogrammetry or LIDAR * 3 √ A 1 † Design phase/ Field QC Review of design with emphasis on application of AMG 4 √ 2 Project start up, Field QC Have current software versions and firmware upgrades 5 √ C D √ 3 Field QC Have back-ups with spare inventory 6 √ 1 Project Start up, Field QC Establish appropriate technology selection guidance criteria based on project tolerance requirements 7 √ 4 Field QC Follow standard operation procedures for maintenance, training, and inspection 8 √ √ 5 √ Field QC Follow standard procedures for maintenance, training, and inspection protocols to check wear parts, and so forth ** § 9 √ 6 B Field QC Training and standardized protocols G § §§ 10 √ 7 †† √ 7 †† Operator assessment, Field QC Training and standard guidelines on QA/QC testing and procedures §§ 11 √ E Project start up, Field QC Establish appropriate technology selection guidance criteria based on project tolerance requirements *** Notes: Error # as defined in Table 9-1. Notes based on workshop participant feedback indicated in frequency columns: 1Frequent although not daily 2Software misuse 3Hardware misuse 4Generally not too many issues 5May need new calibration or a valve change to link with how machine is being used 6Major issue to consider 7Function of project size and application Notes based on field personnel (contractor and workshop participant) experiences indicated in frequency columns: A Owner and contractor not using the same DTM; Cooperator following wrong line on screen; Blade wear incorrect (unchecked for long time); DIntermittent changes in GPS vertical accuracies; ELaser-based systems were influenced by strobe lights during operation; FMost common circumstance leading to reduced accuracy (experience of a contractor); GPeriodic checks (about every 2 weeks) for GPS base station and daily checks of grade behind machines Notes based on project survey outcomes indicated in mitigation strategy columns: * 81% of contractors, 74% of agency personnel, and 87% of heavy equipment manufacturers responded that inaccuracies in the original survey contained in the DTM is one of the major factors contributing to the overall AMG Accuracy. ** 77% of contractors, 62% of agency personnel, and 63% of heavy equipment manufacturers responded that machine response time to positioning information (hydraulic control response) is a major factor contributing to the overall AMG Accuracy. *** 77% of contractors, 75% of agency personnel, and 81% of heavy equipment manufacturers responded that limitations in the positioning methods are one of the major factors contributing to the overall AMG accuracy. § 80% of contractors, 75% of agency personnel, and 81% of heavy equipment manufacturers responded that lack of operator training in heavy equipment operation is one of the major factors contributing to the overall AMG accuracy. §§ 52 to 100% of contractors, 70 to 90% of agency personnel, and 60 to 100% of heavy equipment manufacturers responded that training/competencies of model builders, field inspectors, and grading machine operators is one of the major factors contributing to the overall AMG accuracy. † 77% of contractors, 65% of agency personnel, and 63% of heavy equipment manufacturers responded that end-user misuse of products (equipment, hardware, and software) is one of the major factors contributing to the overall AMG accuracy. †† 88% of contractors, 74% of agency personnel, and 88% of heavy equipment manufacturers responded that failure to identify inaccuracies during the QA/QC process is one of the major factors contributing to the overall AMG accuracy.

NCHRP Project 10-77 98 A frequent problem reported with GPS-based technologies is limited availability of satellites (and, consequently, poor signal attenuation) when operating close to structures, trees, or underground environments. Currently, the U.S. Air Force is committed to maintaining availability of 24 operational GPS satellites, 95% of the time (U.S. Air Force 2014) and is projecting for increased number of satellites in the future. Hein et al. (2007) indicated that the relative gain in accuracy from an increased number of satellites will be marginal. However, AMG users can expect to increase the chances of having the minimum number of satellites required to achieve a certain amount of accuracy because of the new additional satellites. Inertial navigation systems (INS) use gyroscopes and accelerometers and can be used to provide continuous position, velocity, and altitude during only a short signal outage, as the maximum signal outage time is very limited (Mautz, 2008). Recent advancements with use of HA-NDGPS with initiatives from FHWA, globally positioned GDGPS and International GNSS Service (IGS) technologies is providing opportunities to achieve cm level accuracy without significant on-site investment. U.S. Air Force is currently in the process of developing and launching a next-generation GPS satellite (GPS III) which will be available for all military and civilian applications with improved accuracies (U.S. Air Force 2014). Barnes et al. (2003) described LocataNet, a pseudolite-based positioning technology (www.locatacorp.com), which consists of a network of terrestrially-based and time-synchronized pseudolite transceivers that transmit GPS-like signals to obtain position measurements. The signals transmitted using pseudolites are several orders of magnitude stronger than the signals transmitted via GPS; therefore, they are less affected by nearby structures, trees, and so forth. (Mautz, 2008). Based on a kinematic performance test on Locata, Barnes et al. (2003) reported that about 80% of the values are accurate within ±20 mm. LocataNet can be used to augment GPS measurements where necessary, although, theoretically, with enough pseudolites, it is possible to replace GPS (Barnes et al., 2003). GPS with laser or ultrasonic augmentation offers improved vertical accuracies (2 to 6 mm) (Trimble, 2008). From recent field studies on concrete paving projects in Iowa, Cable et al. (2009) found that laser-augmented GPS measurements are somewhat capable of guiding the paver and controlling elevation to achieve a reasonable profile for low-volume roads, but recommended that improvements (or fine tuning) in software is required to better control the elevation that will result in smoother surface profiles. Table 9-4 summarizes a few AMG and equipment application categories and the associated vertical and horizontal accuracy requirements.

NCHRP Project 10-77 99 Table 9-3. Summary of Different Position Measurement Technologies System Accuracy Range User Cost Reference Conventional GPS (no corrections) Variable, > 5 m Global Low DoD, 2008 Assisted GPS (via mobile phones) Variable, 2 to 10 m Global Low Mautz, 2008 GPS integrated with INS Variable Global Variable Mautz, 2008 WAAS or Satellite Based Augmentation System (SBAS) 1.6 to 3.2 m horizontal and 4 to 6 m vertical Global Low FAA, 2008 Nationwide differential GPS (NDGPS) 1 m within 150 km of the broadcast site Global Low ARINC Inc., 2008 HA–NDGPS 10 cm horizontal and 20 cm vertical Global Low – currently in development FRP, 2012 Global DGPS 10 cm horizontal Global Low NASA, 2014 IGS <10 cm horizontal and vertical Global Low Moore, 2007 RTK GPS cm* Global Moderate to high Mautz, 2008 Locata (pseudolites) 6 mm 2 to 3 km High Barnes et al., 2003 Laser- augmented GPS 3 to 6 mm 300 m/line of site radius of laser source Moderate to high Trimble, 2008 Laser ±2 mm Low to moderate Retscher, 2002 Robotic total station ±2 mm 700 m/line of site radius of source High Retscher, 2002 Ultrasonic ±1 mm Immediate reference Low to moderate Trimble, 2008 Ultrasonic augmented GPS ±1 mm Immediate reference Moderate to high Trimble, 2008 Infrared laser 0.1 to 0.2 mm 2 to 80 m High Kraut-Schneider, 2006 * About 90% of survey respondents reported horizontal accuracy of 2 cm or less and 45% of respondents reported vertical accuracy of 2 cm or less with GPS (See Chapter 3) Signal outage or poor satellite reception problems with GPS have been addressed using mobile phones, INS, and pseudolite-based technologies. Use of mobile phones is referred to as Assisted GPS, wherein mobile phones provide information of the satellite Ephemeris, Almanac, differential corrections, and other relevant information (Mautz, 2008). However, the level of accuracy achieved is variable and relatively poor (±10 m), compared to RTK GPS. GPS integrated with

NCHRP Project 10-77 100 Table 9-4. Comparison of different types of construction machines (after Retscher 2002) Equipment Major Application Field* Typical Precision Requirements Dozer Bulk earthworks and earthmoving up to ± 20 mm Grader Fine grading, side slope work up to ± 5 mm Road Paving machine Asphalt/concrete surfaces for pavements up to ± 5 mm in plane and ± 3 mm in height Slip form paving machine Concrete surface for pavements and high-speed railways up to ± 5 mm in plane and ± 2 mm in height *Vertical accuracy requirements (Houghton, 2001): finished surface: < ± 6 mm; base course: ± 6 mm; upper road surface: ± 8 mm; road base: ± 15 mm; Subbase: ± 10 to 30 mm; formation and cap: ± 20 to 30 mm. Accuracy requirements (Peyert et al. 2000): subbase: ± 30 mm; base: ± 20 mm; binder course: ± 15 mm; wearing course: ± 5 mm Construction Process and Human Errors The overall accuracy of the AMG process includes these construction process parameters: • Speed of operation • Direction of travel • Terrain • Material type and support conditions (uniformity) These parameters have not been thoroughly studied or documented in the technical literature and they are application-specific or machine-specific. A brief explanation of each is presented below and a statistical approach to quantify the influence of these factors on the overall accuracy of the AMG process is presented in the following section of this report along with some experimental test results. The level of impact for each of these factors differs with the application type. Speed of operation affects AMG accuracy and overall project costs. Increasing speed decreases the ability of machines to react to error signals and, consequently, reduces the accuracy of the measurement. However, productivity declines as speed declines, impacting project costs. The effect of speed of operation is clearly interlinked with the abilities of the position measurement technology feedback response time. The terrain on a job site can have an impact. Although not critical for paving and fine grading applications, terrain can be critical for general earthwork and excavation applications. The type of material and support conditions under the equipment (whether stable or unstable, uniform or non-uniform) impacts the overall accuracy. Unstable or non-uniform support conditions under the equipment make it more difficult to maintain control relative to the reference. This factor can play a critical role in paving and fine grading applications, and may not be as critical for general earthwork and excavation applications. A conceptual illustration of resulting pavement layer thickness with and without uniform support conditions is presented in Figure 9-1.

NCHRP Project 10-77 101 Pr ob ab ili ty D en sit y Pavement Layer Thickness Desired Target Deviation of mean from target Non-Uniform Support Conditions Uniform support conditions Figure 9-1. Conceptual Illustration of Comparison between Pavement Layer Thickness with Uniform Support Conditions and Non-Uniform Support Conditions Using AMG AMG Control for Single Phase versus Multiple Phases of Project Evaluations of accuracy are primarily based on assessment of one phase of the project or one machine. According to Jack Maclean in a personal communication in 2010, a potential limitation when applying GPS control to highway work is using it as a control for a single phase of the work. Further, the preparation (or analysis of the existing pavement profile and design of the new and milled or trimmed surface) is the most key step. Once the analysis and prep work is completed properly, different systems of paver control have the potential to produce superior results. This concept of building on layers of materials placed with AMG control need to be further investigated. QUANTITATIVE EVALUATION OF ACCURACY Statistical Data Analyses Approach The statistical analyses described herein are to evaluate the accuracy and the precision achieved for an application using different position measurement technologies. The term precision parameter is important to address here, as it relates to the repeatability and reproducibility of the process. Repeatability refers to the variation in repeated measurements made on the same subject under identical conditions (Taylor and Kuyatt, 1994). This explains the variations observed in measurements made using the same setup, equipment, operator, and method, over which there is no change in the measuring property. Reproducibility refers to the variation in repeated measurements on the same subject under changing conditions (Taylor and Kuyatt, 1994). The changing conditions may be due to a change in operators, material types, support conditions, direction of machine travel, etc. Equations 9-1 through 9-3 provide the formulae involved in determining the repeatability and reproducibility errors. The reproducibility variation is evaluated separately for each changing condition, first, and then is combined in the end to calculate the overall effect. MSEityrepeatabil =σ (9-1)    − − +=σ m MSE mI MSAC)1I( mI MSC,0maxilityreproducib (9-2)

NCHRP Project 10-77 102 where m repeated measurements obtained at a given location for I number of times by J number of operators or material types or support conditions or travel direction, using estimates from two-way analysis of variance (ANOVA) results (see Table 9-5). 2 )etc,material,operator(ilityreproducib 2 ityrepeatabilR&R σ+σ=σ (9-3) Table 9-5. Typical Two-Way ANOVA Table Source SS (sum of square) DOF (degree of freedom) MS (mean square) Location (I) SSA I-1 MSA = SSA/(I-1) Number of operators or operating conditions or support conditions etc. (J) SSC J-1 MSC = SSA/(J-1) I x J (interaction term) SSAC (I-1) (J-1) MSAC = SSAC/(I-1) (J-1) Error SSC IJ (m-1) MSE = SSE/IJ(m-1) Total SSTot IJm - 1 — To quantitatively differentiate the capability of each measurement technology specific to the application, the gauge capability ratio (GCR) factor can be computed using equation 9-4. The lower the GCR, the better the capability of the measurement technology for that application. LU 6GCR R&R − σ = (9-4) where U and L are the upper and lower tolerance limits, respectively. Experimental Test Results Field experiments were conducted on three project sites in 2014 on GPS devices mounted on roller compactors. The objective of the field experiments was to demonstrate the use of the statistical analysis approach outlined above to assess the accuracy of the vertical (elevation) measurements from GPS with and without the influence of travel direction. The machines were operated in 3 to 4 km/h nominal speeds and the data was collected approximately every 0.2 to 0.3 m. Project 1 consisted of a Caterpillar CS74 roller with a padfoot shell kit and a RTK-GPS over an earthwork grading project with sloping terrain (Figure 9-2). Machine was operated using 5 passes each in up slope and down slope directions. The elevation difference was about 4 m over the 100-m long test strip. Project 2 consisted of a Caterpillar CS56 smooth drum roller with RTK-GPS over a relatively flat ground of a gravel road (Figure 9-3). Machine was operated using 16 passes in one direction. Project 3 consisted of a CS74 smooth drum roller with GPS receiving satellite-based augmentation system (SBAS) corrections (no RTK) over a sloping gravel road (Figure 9-4). Machine was operated using 4 passes each in up slope and down slope directions. The elevation difference was about 1.5 to 2 m over the 150-m long test strip. Results from multiple passes from each site are shown in Figure 9-5. Repeatability measurement error analysis results are summarized in Table 9-6, and R&R analysis results analyzing the influence of change in direction of travel are summarized in Table 9-7. In comparison with the vertical accuracy requirements summarized in Table 9-5 for bulk earthwork construction, the results indicate that machines equipped with RTK-GPS system on projects 1 and 2 can meet the tolerances while non-RTK GPS used

NCHRP Project 10-77 103 on Project 3 cannot. While change in travel can increase the reproducibility error, having RTK-GPS can help significantly minimize the error. Figure 9-2. CS74 roller setup with padfoot shell kit and RTK-GPS on Project 1 with sloping uneven ground conditions a grading project Figure 9-3. CS563E smooth drum roller setup with RTK-GPS on Project 2 with relatively flat grade over gravel road

NCHRP Project 10-77 104 Figure 9-4. CS74 smooth drum roller setup with SBAS-GPS on Project 3 with sloping gravel road Assessment Plan The research team developed an evaluation matrix of the several factors discussed above (machine related, technology related, and human related), which contribute to the overall accuracy of the AMG process. The evaluation matrix is organized by separating the application categories as follows: • General earthwork • Fine grading • Excavation • Paving The result of the evaluation matrix is a table providing a qualitative assessment of each of these factors, in terms of its level of significance to the overall accuracy of the AMG process (low (L), medium (M), or high (H)) for each application category, based on quantitative information (See Table 9-4). As an example, Table 9-8 was prepared with a preliminary assessment for the influence of several factors on several applications. This assessment approach was applied to the application for intelligent compaction position for two different rollers at three different sites (described earlier). Future studies that desire to quantify the accuracy of AMG technologies with experimental testing and expand the matrix are encouraged.

NCHRP Project 10-77 105 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 E le va tio n (m ) 257 258 259 260 261 262 263 264 Pass 1 - Downslope Pass 9 - Downslope Pass 2 - Upslope Pass 10 - Upslope Measurement Error based on one direction = < 1 cm Measurement Error based on both directions = 2.4 cm Project 1: CS74 with padfoot shell kit and RTK-GPS 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 E le va tio n (m ) 150 151 152 153 154 155 156 157 Pass 1 Pass 5 Pass 15 Measurement Error based on one direction = < 1 cm Project 2: CS56 smooth drum with RTK-GPS Distance (m) 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 E le va tio n (m ) 349 350 351 352 353 354 355 356 Pass 5 - Up Slope Pass 11 - Up Slope Pass 6 - Down Slope Pass 12 - Down Slope Measurement Error based on one direction = 4.6 cm Measurement Error based on both directions = 6.7 cm Project 3: CS74 smooth drum without RTK-GPS (SBAS) Figure 9-5. CS56 smooth drum roller setup with RTK-GPS on Project 2 with relatively flat grade over gravel road

NCHRP Project 10-77 106 Table 9-6. Summary of repeatability analysis results on GPS elevation measurements on three earthwork project sites with and without RTK-GPS mounted on roller compactors Project Direction Elevation Range (m) Average (m) Measurement Error (cm) Project 1: RTK-GPS mounted on CS74 roller with padfoot shell kit Down Slope (5 passes) 258.7 to 262.6 260.8 0.9 Up Slope (5 passes) 0.9 Both Directions (10 passes) 2.4* Project 2: RTK-GPS mounted on CS563E smooth drum roller Relatively flat grade (16 passes) 153.0 to 153.3 153.2 0.6 Project 3: SBAS (no RTK) on CS74 smooth drum roller Up Slope (4 passes) 351.2 to 352.9 352.3 4.6 Down Slope (4 passes) 4.0 Both Directions (8 passes) 6.7* *Data includes both directions – see impact of change in direction of travel in R&R analysis results Table 9-7. Summary of R&R analysis results on GPS elevation measurements to assess the influence of change in direction of travel Project Elevation (cm) Percent contribution* of σreproducibility Impact of change in direction on measurement valuesσrepeatability σreproducibility σR&R Project 1: RTK-GPS mounted on CS74 roller with padfoot shell kit 1.2 3.0 3.4 89.4 Significant Project 3: SBAS (no RTK) on CS74 smooth drum roller 11.0 13.1 17.1 59.8 Significant *100 x σ2repeatability /σ2R&R

NCHRP Project 10-77 107 Table 9-8. Example Qualitative Assessment matrix of Accuracy Factors for Various Application Categories C at eg or y Machine Type Machine Related Errors Technology Related Errors Human Error M at er ia l T yp e Su pp or t C on di ti on s Sp ee d of O pe ra ti on T er ra in Sl ip L ow A cc ur ac y G P S (n o co rr .) H ig h. A cc ur ac y G P S (w it h co rr .) L as er A ug . G P S U lt ra so ni c A ug . G P S So na r R T S G en er al E ar th w or k Scraper H M M M H M L L L M L M Dozer M L M M M H L L L H L M Pulverizer M M H M M H L L L H L M Moisture/ Chemical Control H M H M M H L L L H L M Grader M H H M M H L L L H L M Fi ne G ra di ng Grader M H H M M H L L L H L M Trimmer M H H M M H L L L H L M Ex ca va tio n Excavator H M H M M H L L L H L H Loader H M H M M H L L L H L H Pa vi ng Miller M H H M M H L L L H L M Concrete M H H M M H L L L H L M Asphalt M H H M M H L L L H L M L, M, and H indicate low, medium, or high significance ACCURACY OF AMG SUMMARY • AMG component accuracies is an issue that affects various stages of the process including: Initial data collection for developing existing surface terrains; development of DTM and EED, AMG processes, procedures, and end-user competencies, QA/QC reported practices, heavy and fine grading equipment operations, and paving equipment operations.

NCHRP Project 10-77 108 • Sources of errors contributing to the overall AMG accuracy were identified as an outcome from the AMG Phase I Workshop and were related to frequency of error, means for detection, and mitigation strategies to overcome the error. Survey outcomes from contractors, vendors, and agency personnel on factors contributing to the overall accuracy of AMG are cross- referenced to the findings from the workshop results. It does not appear that the effect of construction process and human errors has ever been thoroughly studied or quantified. Most contractors, vendors, and agency personnel who responded to the survey questions reported that these variables play a key role in the overall accuracy of the AMG process. • A frequent problem reported with GPS-based technologies is limited availability of satellites (and, consequently, poor signal attenuation) when operating close to structures, trees, or underground environments. Currently, the U.S. Air Force is committed to maintaining availability of 24 operational GPS satellites, 95% of the time (U.S. Air Force 2014) and is projecting for increased number of satellites in the future. While the relative gain in accuracy from an increased number of satellites will be marginal (Hein et al., 2007), AMG users can expect to increase the chances of having the minimum number of satellites required to achieve a certain amount of accuracy because of the new additional satellites. • The overall accuracy of the AMG process includes these construction process parameters: speed of operation, material type and support conditions (uniformity), and terrain. These parameters have not been thoroughly studied or documented in the technical literature and they are application-specific or machine-specific. A statistical analysis approach to quantitatively assess the influence of these different parameters is presented along with some experimental results. The level of impact for each of these factors differs with the application type. • A detailed evaluation of the several factors (machine related, technology related, and human related), which contribute to the overall accuracy of the AMG process, are encouraged for future study. An assessment matrix is provided to rate factors by significance.