Study: Comparison Between Photogrammetry and Laser Radar

At the Harris Corp. (Government Communications Systems Division—GCSD), solid carbon-fiber and aluminum-shaped reflector dishes are manufactured to support military ground, sea, and air communications. These reflectors range in size from 1 to 13 m. To meet performance requirements, the surface profile of these reflectors are held to rigid tolerances. For 25 years, photogrammetry (PG) has been the Harris standard for characterizing the surface profile of these reflectors due to the technology’s high accuracy, portability, and ability to work in unstable conditions. The technique has been “tried and true” and is familiar to engineering and quality organizations.

During the past year, laser radar (LR) has been evolving into a robust inspection technique that reduces materials and automates the measurement process. In a cost-competitive market of reflector fabrication, laser radar offers a promising solution in reducing the recurring labor cost during significant production efforts. To implement this technology, it was necessary for the LR to be compared to what Harris recognizes to be the standard of surface characterization. To be accepted, the surface profile characterization (i.e., RMS) for reflectors must be less or equal to 10 percent when compared to photogrammetry.

This article discusses the LR evaluation techniques used for determining the surface profile of reflectors and whether LR would be a suitable alternative to PG during significant production efforts at Harris.

On a variety of reflector sizes, the surface profile (i.e., RMS) was calculated with both LR and PG. The percent differences between RMS values ranged from 2 to 6 percent. Additionally, the labor hours required to determine the RMS value were reduced 90 percent when an automated program was used. For this application of determining the surface profile of reflectors, LR was a suitable alternative to PG in reducing recurring costs associated with production efforts.

Future investigation will be performed to utilize a target projection system in conjunction with PG. If successful, the target projector would eliminate the need to apply and remove PG targets and greatly reduce measurement time.

Surface profile characterization

Please note the process and technique used to generate the RMS values for the different methods are Harris-specific; other applications may give different results.

A symmetrical point grid is determined so that the reflector’s surface profile can be calculated. This grid is very labor intensive and must be applied accurately because it will be used as an alignment feature during the surface characterization. The grid is applied in a repeatable sequence and correlates the datum reference points of the production tool. The grid is typically transferred to the surface of the reflector during production. This repeatability is maintained throughout production to achieve an “apples to apples” comparison among the reflectors. After measurement, the data are compared to the theoretical surface, and new Z values are generated for the measured X and Y values. Next, the theoretical Z values are used to generate a deviation, or delta, between the measured and the intended surface. These data are then used for reporting RMS and characterizing the surface profile of the reflector. A requirement of less or equal to 10 percent was determined as the threshold needed for the RMS comparisons.

PG and LR comparison process

To ensure that the study comparison was not jeopardized, it was decided that the reflector would be put into alignment and measured with the LR first. A LR RMS value was then calculated using the predetermined measurement grid. The LR was used immediately afterward to project the point grid onto the surface, which allowed for easy application of PG targets. This method ensured that the same point locations are consistent between both measurement methods. A PG RMS value was then calculated using this applied target grid. Because divulging RMS values is competition-sensitive, the LR and PG RMS will be expressed as a percentage difference. The formula below is used:

It is important to note that comparing the calculated PG and LR RMS is inadequate for determining the relative accuracy of the two technologies because the surface of the reflector is less accurate than either method. It is, however, sufficient for determining whether LR is accurate enough for the production process.

1.3-m reflector

For a 1.3-m reflector (see figure 1), comparison data were captured on serial numbers 16, 17, and 18 during a recent manufacturing build. The comparison study mimicked the 88-point grid used by PG to verify reflector RMS throughout the manufacturing process. Plots were generated from the Z deviations to aid in graphically monitoring the surface of the reflectors. The tracking of peak-to-peak Z values between the two methods was consistent (see figures 2, 3, 4). For serial numbers 16, 17, and 18, the percent difference in RMS values were 5.8 percent, 5.9 percent, and 6.2 percent, respectively.

Figure 1: 1.3-m reflector

 

Figure 2: 1.3-m reflector, SN 16

 

Figure 3: 1.3-m reflector, SN 17

 

Figure 4: 1.3-m reflector, SN 18

 

1.3-m reflector, 288 grid

To evaluate the LR sensitivity of a denser point grid (288 points), another 1.3-m reflector was used. The denser grid increased the resolution and helped minimize the deviations for RMS and peak-to-peak values. The tracking of peak-to-peak Z values between the two methods was consistent (see figure 5). For this particular reflector, the difference between LR and PG RMS values was 5.6 percent.

Figure 5: 1.3-m reflector, denser grid

 

2.5-m reflector

The second size chosen for comparison was a 2.5-m panelized reflector (see figure 6). This particular reflector is larger and deeper than the smaller 1.3-m reflector. This reflector was characterized using 320 data points, which represents the current PG process. The same measurement technique, alignment process, and analysis used during the 1.3-m tests were applied to ensure consistency between data sets. The tracking of peak-to-peak Z values between the two methods was consistent (see figure 7). For this particular reflector, the difference between LR and PG RMS values was 1.6 percent.

Figure 6: 2.5-m panelized reflector

 

Figure 7: 2.5-m reflector

 

3.8 meter reflector

A third data point was a 3.8-m trailer-based, panelized antenna (see figure 8). This larger reflector was characterized using 500 data points. Again, the same measurement technique, alignment process, and analysis used in the 1.3-m test and the 2.5-m test were applied to ensure consistency between data sets. The tracking of peak-to-peak Z values between the two methods was consistent (see figure 9). For this particular reflector, the difference between LR and PG RMS values was 2.6 percent.

Figure 8: 3.8-m trailer-based, panelized antenna

Figure 9: 3.8-m antenna

Labor savings

A labor-saving analysis was performed on the 2.5-m reflector. The results below are typical, but might vary depending on the size of the reflector. For PG and LR, the following represented the four process steps that were compared:

Photogrammetry

• Setup—Target grid application

• Measurement—Perform photo shoot

• Process data—Generate report and plot

• Cleanup—Remove targets and clean reflector surface

 

Laser radar

• Setup—Position LR and perform reflector alignment

• Measurement—Perform scan

• Process data—Generate report and plot

• Cleanup—Stow LR

 

The labor savings is displayed as a relative comparison between PG and LR. The formula below is used:

In figure 10 below, ~ 90-percent time reduction was realized with the use of the LR.

Figure 10: Labor comparison

 

Most of the PG time was spent applying target, removing targets, and processing data in accordance with Harris’ standard procedures. But data processing time for PG could be greatly reduced via the use of newly developed scripting techniques.

The findings showed that it is possible to perform 100-percent unit inspection with LR and reduce recurring costs compared to the lower rate-sampling approach used with PG. For this application of determining the surface profile of reflectors, laser radar was a suitable alternative to photogrammetry in reducing recurring labor associated with a production environment.

Conclusion

For small to medium-size reflectors in a production environment, the use of the LR significantly reduced the recurring labor required to characterize the surface profile (i.e., RMS) of reflectors. This evaluation demonstrated that when compared to PG, LR met the requirement of less or equal to 10 percent from a dimensional comparison standpoint. In addition to achieving parity in measurement performance to PG, LR achieved significant cycle-time reduction with the use of automated programs. The difference in RMS was 6 percent maximum for small reflectors and substantially smaller for the larger 2.5-m and 3.8-m reflectors. Peak-to-peak Z values measured by LR were consistent when compared to PG. The comparison study satisfied the requirements needed to implement LR in areas of heavy production where cycle time and labor costs are driving factors.

Further investigation is expected using PG and a newly developed target projector, which may further reduce cycle time and labor costs.

About The Author