A Comparison of Methods for Assessing Space Suit Joint Ranges of Motion Lindsay T. Aitchison1 NASA Johnson Space Center, Houston, TX, 77586 Through the Advanced Exploration Systems (AES) Program, NASA is attempting to use the vast collection of space suit mobility data from 50 years worth of space suit testing to build predictive analysis tools to aid in early architecture decisions for future missions and exploration programs. However, the design engineers must first understand if and how data generated by different methodologies can be compared directly and used in an essentially interchangeable manner. To address this question, the isolated joint range of motion data from two different test series were compared. Both data sets were generated from participants wearing the Mark III Space Suit Technology Demonstrator (MK-III), Waist Entry I-suit (WEI), and minimal clothing. Additionally the two tests shared a common test subject that allowed for within subject comparisons of the methods that greatly reduced the number of variables in play. The tests varied in their methodologies: the Space Suit Comparative Technologies Evaluation used 2-D photogrammetry to analyze isolated ranges of motion while the Constellation space suit benchmarking and requirements development used 3-D motion capture to evaluate both isolated and functional joint ranges of motion. The isolated data from both test series were compared graphically, as percent differences, and by simple statistical analysis. The results indicated that while the methods generate results that are statistically the same (significance level p= 0.01), the differences are significant enough in the practical sense to make direct comparisons ill advised. The concluding recommendations propose direction for how to bridge the data gaps and address future mobility data collection to allow for backward compatibility. Nomenclature deg = degrees in. = inches I. Introduction S pace suit design is rooted in aviation history with the first practical American garment arriving on the scene in 1934, as a result of partnership between the aviator Wiley Post and the B.F. Goodrich Company (Thomas & McMann, 2006). The boom of the of space suit design however rests solidly in the 1960s as engineers and scientists across the United States worked to create the suits that would take American astronauts into space and ultimately to the moon with hopes of Mars. As each suit prototype was fabricated, both the manufacturers and NASA invested significant time to benchmark suit mobility during the completion of representative extra-vehicular activity (EVA) tasks. As the complexity of planned EVA tasks increased from Gemini to Apollo to Space Shuttle and International Space Station Programs, the importance of designing suits for mobility became more obvious, but the methods for evaluating suit mobility remained some what rudimentary relying on hand measurements with goniometers and/or a protractor and still photographs with results often being highly dependent upon the consistency of the person actually taking the measures for identifying and aligning the joint axes to be compared. During the beginnings of the Constellation Program, NASA shifted gears to 3D motion analysis aided by the use of the Vicon motion capture system. The hope was to collect dynamic range of motion data that would be more representative of the entire range each suit could achieve when not bound to hold static poses for traditional still photography. Now, with the transition from the short lived Constellation Program to the decidedly more technology development focus of the Advanced Exploration Systems Program, NASA is attempting to use the vast collection of mobility data from 50 1 Insert Job Title, Department Name, Address/Mail Stop, and AIAA Member Grade for first author. 1 American Institute of Aeronautics and Astronautics years worth of space suit testing to build predictive analysis tools to aid in early architecture decisions for future missions and Exploration Programs. To produce a useful tool for predicting human mobility given a specific combination of mobility features, one must of must understand how the base data set was acquired and if and how data generated through different methodologies can be used interchangeably. The following sections of this paper provide an overview of the different methodologies available to collect and analyze mobility data and then focuses on a comparison of data collected for the Space Suit Comparative Technologies Evaluation Report in 2000 (Ross, 2000) and the CSSS benchmarking and requirements development project in 2009 (England, 2010). These two test series are of particular interest because not only do they evaluate the same suits by two different methods, photogrammetry and Vicon, but they also have a test subject in common. Data from the cross-over subject should yield more direct results as complicating factors such as test subject unsuited mobility, suit fit, and experience can be eliminated for within subject comparisons. II. Background There have been numerous studies to quantify human mobility with motivations ranging from understanding physiology to improving athletic performance to the design of apparel and protective clothing. The majority of the studies define mobility in terms joint range of motion (ROM) as measured by the relative movement between body segments. The methods for collecting data range from simplistic hand measurements with a protractor or goniometer to more complex digital data collection and software aided analysis. For the protective clothing studies, which includes space suits, the acceptability of the garment tends to be gauged by comparing test subject ROM in the minimally clothed versus fully suited conditions for both static and dynamic ranges. For the final down selection of the Apollo space suit assembly (SSA) design, Jones reported on an elaborate series of testing that provided weighted scores for suit performance on each subtest in the overall evaluation (Jones, 1966). Some of the more heavily weighted subtests of his testing were mobility evaluations. Jones reported that mobility was measured as functional reach inside a high fidelity Crew Module simulator and three different analyses of isolated joint ROM outside the vehicle mock-up. The isolated data were collected by real-time measurements with a goniometer as subjects held static postures in each joint’s extreme ranges of motion. Additionally, multi- exposure photos of subjects posing in postures demonstrating the extreme ranges of each joint motion in front of a grid board (see Figure 1), were taken and later analyzed to determine angular range (a process alternately termed photogrammetry or strobe analysis). To complete his study, Jones also conducted an x-ray study which used the midline of the major bones to directly measure the human joint angles while wearing the suit prototypes. Figure 1: Example of strobe analysis method for minimally clothed subject (URS Corporation, 1974) 2 American Institute of Aeronautics and Astronautics One can infer that Jones and his team regarded the strobe method as most accurate as its results were weighted twice as much as the other two methods’ results. The strobe method was again employed at the start of the Space Shuttle Program by the URS Corporation in order to evaluate the Apollo A7LB SSA against a new Orbital Extravehicular Spacesuit (OES) for microgravity operations. The conclusions of the URS report indicate that there was significant difficulty in obtaining pure isolated joint movements and that the bulk of the suit led to what the researchers believed was an exaggerated appearance of movement in the photographs. Additionally, the researchers reported that the static poses captured by photography are more representative of the “normal range of motion” than the extreme. This conclusion was attributed to the observation that test subjects were able to over-drive the joints to achieve a greater range but could not maintain the pose long enough to acquire clear still photos (URS Corporation, 1974). More recently, a joint venture between NASA and the Russian Space Agency to evaluate Mars suit concepts in 2004 relied solely on the Apollo era photogrammetry method to evaluate static range of motion for isolated joints. The Mars suit evaluations determined neutral joint positions and angular range by direct measurements taken from still photographs of subjects performing prescribed motions in a plane parallel to the grid board (Abramov, 2005). As digital technology improved over the last quarter of the 20th century, new forms of real-time motion capture analysis emerged. Some of the most common of the optical motion analysis systems are Dartfish, Simi Motion Capture 3D, Ariel Performance Analysis System (APAS), and Vicon MX. Systems like Dartfish and APAS use digital video cameras and computer vision techniques to create a 2D model of the subject without the use of markers (Brock, 2010). The Vicon system, by contrast, uses reflective marker balls to generate 3D models with the BodyBuilder software. The benefit of the 3D model is that it enables the researcher to collect the complete coupled motion of complex joints, such as the hip and shoulder, throughout the entire dynamic range of motion; this reduces the loss of data outside the “normal range” associated with static ROM collected via photogrammetry (Doriot et al, 2006). However, there are several marked disadvantages compared to manual methods including large capture volumes required, time consuming calibration and start-ups steps, sensitivity to camera vibration, long and tedious post-processing, and, of course, the high cost of purchasing the systems. The Dartfish system was employed at the Desert Research and Technology Studies 2003 field test to evaluate the efficacy of the MK-III prototype space suit in the performance of field geology tasks but it yielded extremely varied within subject results that the test team judged to be highly erroneous (Ross, 2004). The Vicon system has been used in several test series at NASA Johnson Space Center’s (JSC) Anthropometrics and Biomechanics Facility (ABF) since being acquired by the ABF team in 2004. The results of the Vicon test series have been used to aid lunar rover design, evaluate the effects of mass and center of gravity on gait patterns, and benchmarking the mobility of current flight and prototype space suit technologies. However, the author was unable to locate any reports comparing the relatively newly collected Vicon data with the decades of previously collected ROM data on the same or similar space suits. The Space Suit Comparative Technologies Evaluation Report, which is detailed in the following section, addressed the validity of digital motion capture using the Ariel motion analysis system (Ariel Dynamics). The general conclusions of that testing revealed slight difference between data collected manually and digitally but were generally felt to be comparable. The Ariel Motion Analysis System is considerably less sophisticated than the Vicon system used to benchmark space suit ROM in the Constellation requirements development effort, thus a comparison between the Vicon system and manual measurements is still critical to understanding the extent to which data collected by the two different methods may be used interchangeably. A. 1999 Prototype Suit Mobility Study – Photo Method In 1999, JSC’s space suit engineering and human factors teams, led by Amy Ross and Dr. Sudhakar Rajulu, respectively, collaborated on an extensive test series to evaluate the performance capabilities of candidate prototype suits with the aim of collecting both subjective and objective data regarding the performance of various planetary space suit mobility joints to aid in the downselction of a mobility system for the next generation space suit assembly (Ross, 2000). One subset of the overall series was the evaluation of suited range of motion during the performance of both isolated tasks. Thirteen isolated movements were performed by each of three test subjects in shirtsleeves and both the WEI and MK-III suits. The motions were performed in an identical manner by each subject to reduce variability in the results. The team captured data both by still photography and digital video recording, but only the still photography method will be presented in detail and discussed here. The photographic data collecting utilized two 6in. x 6in.-gridded boards, one vertical against the wall and the other horizontal on the floor, and two still cameras. One of the cameras maintained a fixed location orthogonal to the vertical grid board throughout the test; second was fixed in position on the ceiling directly above the center of the horizontal grid board. A photo was taken at each extreme that could be achieved and held reasonably stable by the subjects as the isolated motions were performed. 3 American Institute of Aeronautics and Astronautics Post-test analysis of the still photographs was completed by printing the photographs onto transparency films, laying the photos of the two extremes of a motion on top of each other on a light table, and measuring the angle difference using a protractor. The method required the analyst to be familiar with the mechanics of the suit in order to properly assume locations for neutral suit positions, joint centers of rotation, and the joint segment centerlines. The results for each individual test subject in each test condition were tabulated to illustrate the total joint range achieved and also used to calculate overall percent differences in mobility when going from the shirtsleeves to suited condition. Lastly, the reported results indicate “smallest effective range” for each suit defined as the smallest maximum range of motion produced by the three test subjects during the performance of isolated joint movements for each suit (Ross, 2000). In addition to the quantitative results, Ross provided a detailed summary of test subject comments and test conductor observations of each test condition. These subjective comments are valuable in understanding contributing factors when the reported angular data for an individual or motion seem out of context when compared to other suited or shirtsleeves conditions. The test conduct comments also aided the present author in understanding how the original test conductor established joint centers and their assumptions of what constituted an ‘isolated’ motion. B. 2008 Constellation Suit Requirements Development Study – Vicon Method With the advent of the Constellation Program (CxP), the NASA Constellation Space Suit System (CSSS) Pressure Garment Team partnered with the ABF to develop a mobility test methodology that would enable rapid benchmarking between suit prototypes, verifiable range of motion requirements for CxP suits, and computer modeling of suit system behavior in a consistent and repeatable fashion. The ABF selected the Vicon 612/SV motion capture system to record the data and the complementary Vicon BodyBuilder™ software to analyze the data. The Vicon set-up used 10 cameras positioned over the capture volume to record the reflections of the 41 marker balls which were fixed to the joint segments of the test subject (see Figure 2). Figure 2. Pressurized WEI (left) and MK-III suits with marker ball set in place There were a minimum of three markers per body segment on the torso and right hand side limbs that roughly corresponded to biomechanical human body segments; the symmetry assumption was used to greatly reduce data reduction times in post-processing. The same marker sets were used for all test subjects in all conditions. A total of four test subjects (2 female, 2 male) completed a series of 16 isolated and 49 functional tasks representative of CxP mission objectives in the unsuited, WEI, and MK-III conditions. The isolated tasks were standardized to ensure comparable data while the functional tasks (e.g. crawling, walking, kneeling, etc.) were left to test subjects to decide on the best way to complete the task in order to capture outliers. Only the isolated task data were used for comparisons in this paper. The BodyBuilder™ software calculated joint angles by assuming a primary axis extending from the dynamic and static joint segments and measuring the angular travel of the dynamic segment axis relative to its neutral position and the static segment’s axis. These measures are illustrated in Table 2, below, along with the names of the two joint segments used to calculate the angle. All of the isolated tasks presented in this paper 4 American Institute of Aeronautics and Astronautics were collected without the use of props against which subjects could gain force advantage (e.g. pressing palm against the wall to increase range of wrist flexion/extension). Table 2: Joint Motions Performed in Digital Data Collection (England, 2010) Joint Motion Illustration Dynamic Static Segment Segment Shoulder Upper arm Torso flexion/extension Shoulder Upper arm Torso adduction/abduction Shoulder lateral/medial Upper arm Torso rotation Elbow flexion/extension Forearm Upper arm Hip flexion/extension Hip Torso Hip adduction/abduction Hip Torso 5 American Institute of Aeronautics and Astronautics Hip (ankle) rotation Hip Torso Torso lean Torso Hip Torso flexion/extension Torso Hip Torso rotation Torso Hip Knee flexion/extension Shin Hip Ankle flexion/extension Foot Shin In his discussion, England emphasizes that suit mobility measurements “must reflect the fact that altered movement strategies are utilized while wearing a space suit” thus while the ranges of motion achieved in each suit conditioned varied, their ability to functionally perform tasks was unhindered on the whole. He further concludes that the 3D Vicon methodology pioneered by his team will yield both time and cost savings in the evaluation and benchmarking of future prototype suits. III. Data Comparisons As indicated in Tables 1 and 2, both test series used the same isolated joint movements to define the ranges of motion, however the signs associated with some joints differ. To minimize confusion, the data were compared as total ranges of motion. The total range for each joint was calculated as the sum of the absolute values of the joint extremes in each direction. For all results shown below, ‘overall’ refers to the singular value listed as a suit’s specific isolate ROM within the two test reports. In both cases, the range of motion was defined as the minimum of 6 American Institute of Aeronautics and Astronautics the total maximum ranges achieved by the subjects for the joint of interest. The ‘individual’ results refer to the ROM data reported for the one test subject who participated in both test series. The results of the ROM data collected by both manual and digital methods are shown in the comparative column graphs of Figures 4 – 6 for all suit conditions. Data was recorded for all motions in all conditions, thus the absence of a column for any motion indicates 0° total range was achieved. Unsuited Isolated Joint ROM 280 260 240 220 200 es)180 e egr160 Vicon Overall e (d140 Photo Overall ng Vicon Individual a R120 al Photo Individual Tot100 80 60 40 20 0 Ankle Ankle Eblow Hip Hip Knee Knee Shldr Shldr Shldr Torso Torso Torso flex/exrotationflex/ex flex/exabd/addflex/ex flex/ex*abd/ad flex/ex lat/med lean flex/exrotation Joint Motion Figure 4. Comparison of digitally and manually collected unsuited ROM data MK-III Isolated Joint ROM 280 260 240 220 200 es)180 e egr160 Vicon Overall e (d140 Photo Overall ng Vicon Individual a R120 al Photo Individual Tot100 80 60 40 20 0 Ankle Ankle Eblow Hip Hip Knee Knee Shldr Shldr Shldr Torso Torso Torso flex/exrotationflex/ex flex/exabd/addflex/ex flex/ex*abd/ad flex/ex lat/med lean flex/exrotation Joint Motion 7 American Institute of Aeronautics and Astronautics Figure 5. Comparison of digitally and manually collected MK-III ROM data WEI Isolated Joint ROM 280 260 240 220 200 es)180 e egr160 Vicon Overall e (d140 Photo Overall ng Vicon Individual Ra120 al Photo Individual ot100 T 80 60 40 20 0 Ankle Ankle Eblow Hip Hip Knee Knee Shldr Shldr Shldr Torso Torso Torso flex/exrotationflex/ex flex/exabd/addflex/ex flex/ex*abd/ad flex/ex lat/med lean flex/exrotation Joint Motion Figure 6. Comparison of digitally and manually collected WEI ROM data The percent difference between the digitally and manually collect ROM data were calculated as (range_digital – range_manual)/(range_digital)*100. Figures 7 and 8 show comparative results for all suited conditions and all thirteen isolated motions identified in the manual data collection test plan. Negative values indicate that the ROM recorded by the manual method exceeded that recorded by the digital method. 8 American Institute of Aeronautics and Astronautics % Difference Vicon vs. Photo 100.0 80.0 60.0 40.0 20.0 MK-III Individual %) 0.0 WEI Individual ce ( Unsuited Individual en -20.0 er MK-III Overall Diff -40.0 fAlenxk/leexroAtnaktiloenfElbexlo/ewx fleHxi/pexabdH/iapddflKenxe/eex fleKxn/eeex*aSbhdl/dard fSlehxl/derx laSth/lmderd Tloerasno fTleoxr/seoxroTtoartsioon WEI Overall Unsuited Overall -60.0 -80.0 -100.0 -120.0 -140.0 Joint Motion Figure 7. Comparison of percent difference digital vs. manual ROM results for each suit condition (dashed red line represents level of practical significance) A comparison of means using the Student t test for a one-way analysis of the percent difference results by data collection method showed no statistically significant difference between the methods for p=0.01. A comparison of means using the Student t test for a one-way analysis of the percent difference results by suit condition also showed no statistically significant difference between the conditions for p=0.01. However, when comparing the data against the 20 % practical significance level (meaning all values within 20% difference of each other during requirements verification would be considered the same) established for CSSS requirements the two methods yield practically identical results in less than five test points per test condition for both the overall and individual comparisons. The highlighted cells in Tables 3 indicate values that have exceeded the practical significance level for the percent difference between the Vicon and Photo method results. There is no discernable correlation between which motions achieved practical significance overall versus across the individual results. Table 3. Comparison of % Difference of Vicon and Photo Results at 20% Practical Significance Overall Results Individual Results Direction Unsuited MK-III WEI Unsuited MK-III WEI Ankle flex/ex 30.1 54.5 49.1 3.3 33.3 14.0 Ankle rotation -130.9 15.9 17.2 Eblow flex/ex -54.2 -44.3 -73.8 8.3 -12.5 -36.0 Hip flex/ex 36.8 3.2 16.7 27.8 -8.1 -32.9 Hip abd/add 35.7 67.3 65.5 19.3 -32.4 31.3 Knee flex/ex 14.2 21.2 2.4 10.7 27.4 -2.0 Knee flex/ex* 58.6 60.7 33.3 78.8 72.0 58.2 Shldr abd/ad 7.6 3.1 25.2 25.0 35.1 40.9 Shldr flex/ex -2.5 35.9 40.2 -18.5 37.7 7.8 Shldr lat/med 11.5 30.1 32.6 -36.7 -1.6 -3.7 Torso lean 34.8 56.7 82.5 -28.2 57.5 85.6 Torso flex/ex 23.0 50.3 52.9 19.2 23.5 17.4 Torso rotation -105.8 4.5 36.7 100.0 100.0 100.0 9 American Institute of Aeronautics and Astronautics While there is a statistical equivalence between the Vicon and Photo methods, the compared data cannot be used in a directly interchangeable manner based on quantitative results alone. It’s logical to assume that the small sample size for both methods is a substantial contributor to the large range of values reported as the effective range for each suit overall but the sample size cannot account for the variances within the cross-over subject’s data. Thus, the focus for finding method-based contributing factors focused on the individual results primarily. Table 4 compares the data across suits and methods to highlight which method generated the greater range for each of the motions; as before, comparisons with percent difference less than 20% are considered equal. Table 4. Method Generating Maximum Range for Each Condition (Within Subject) Test Condition Direction Unsuited MK-III WEI Ankle flex/ex Equal Vicon Vicon Eblow flex/ex Equal Equal Photo Hip flex/ex Vicon Equal Photo Hip abd/add Equal Photo Vicon Knee flex/ex Equal Vicon Equal Knee flex/ex* Vicon Vicon Vicon Shldr abd/ad Vicon Vicon Vicon Shldr flex/ex Equal Equal Equal Shldr lat/med Vicon Equal Equal Torso lean Photo Vicon Vicon Torso flex/ex Equal Vicon Vicon # Responses Equal 6 4 3 Vicon 4 6 6 Photo 1 1 2 Across all conditions for the individual, when there is a practical difference between the ranges generated by the different methods, it will most often be Vicon that yields the greater range of motion. Looking at the suited results for torso flex/ex and torso lean, it is obvious that the Vicon marker set is directly the cause of the seemingly larger ranges. The Vicon method measured torso flex/ex and lean by comparing the relative motion of the torso plate (located on center chest) to the hip plate (located between the thigh bearing and knee joint) whereas the photo method compared motion of the suit upper torso (relatively same as Vicon location) to the center of the suit brief. Thus the Vicon method accounted for the combined mobility of the hips and waist elements to achieve torso flexion which is considerably more than the capability of the waist element alone- particularly for torso lean because the only torso adduction/abduction features for both suits are located in the hip elements. Differences in the way in which the motions were performed appear to be the cause of the greater ranges generated by the Vicon method for both styles of knee flexion/extension. For the ‘functional’ range (noted by * in Table 4), the photo method collected the range at standing and kneeling static positions. The ‘functional’ measure for Vicon however is not directly assigned to a kneeling task; it could have been the result of any of the 49 functional motions that were performed. The prescribed method for isolated knee flexion/extension was slightly different as well because Vicon ranges were again collected as dynamic sweeps as opposed to static poses and the Vicon method also allowed the test subjects to support themselves by holding onto the back of a chair as they performed the sweeps. The fact that the only practical difference occurred with MK-III, leads one to believe that the increased stability was enough to off-set the weight of the comparatively heavy MK-III and allow the subject to use the full joint range of motion. Hip flexion/extension were performed in an almost identical manner across test methods and were thus ruled out as the cause of the variance between ranges recorded. Interestingly, it is only the WEI suit that shows a significant difference in recorded ranges (only 8% difference between methods on MK-III). When suited, hip flexion/extension is not a pure planar motion; it requires a combination of hip element flex/ex and adduction/abduction which, when viewed from the front, makes the knee push away from the saggital plane as the hip flexes. The space suit community commonly refers to this mix of motions within the performance of an isolated task as ‘programming.’ When using the photo method, the extent to which the knee abducts or adducts is impossible to tell and thus it is quantitatively ignored when the full range is recorded. With the Vicon method, however, the analyst self-selects the degree to which they feel the knee strays from the planar motion and thus the maximum flex/ex is actually measured as the angle between the two resultants generated from the summation of motion in the two planes. The data 10 American Institute of Aeronautics and Astronautics