National Aeronautics and Space Administration
CONTRACT: NAS 9-1891 5
(Ariel Dynamics Inc. Note: The following are excerpts from
the full report.
1. WORK ACCOMPLISHED DURING THIS REPORTING PERIOD
A. Literature Search
A fairly extensive literature search was done on recent progress in areas related to the theoretical components of this project. The search was done manually at the libraries of the Massachusetts Institute of Technology over a period of several days, and identified dozens of articles for immediate use. Current work on limb motor control was compiled, to assist in adapting the existing standard limb biomechanical model to handle the nonstandard conditions. Various newly reported analysis methods for dealing with the experimental data were reviewed, as a start in developing the descriptive model based on the empirical results from the nonstandard conditions. A comparison study of recent publications of global optimization techniques was made, in order to find a better algorithm for the highly nonlinear problem of limb movement with many local minima.
C. Ascension Flock Of Birds
The four-receiver Flock of Birds system, manufactured by Ascension Technology, is used in this project for kinematic data acquisition, and represents one of the major pieces of instrumentation in the study. The Flock of Birds was returned to the manufacturer for updating. The circuit boards were upgraded with the latest engineering changes. The read only memories (ROM) were upgraded to reduce the amount of noise in the signals and improve the sustained data gathering rate. As a courtesy, Ascension also evaluated MOCO's customized control and acquisition software to determine if it was exceeding any of the data collection rates.
After receiving the updated Bird system, MOCO revised the experiment software to adapt to the new communication instructions, and integrated the suggested software changes. A subject from MOCO personnel was instrumented and tested under the new modifications. System performance was improved and was more reliable. The noise was reduced and the sustained gathering rate was maintained at 100 Hz per receiver. Ascension had suggested inserting a delay at the start of the data gathering period, but this became unnecessary, and since it would affect the reaction time calculations, the suggested change was not implemented. A research meeting was held in Phoenix, AZ, August 9-11. The purpose of this meeting was to convene MOCO personnel and consultants involved in this project to discuss the various components. Attendees were Drs. Maulucci and Eckhouse from MOCO; Drs. Penny and Ariel from Computerized Biomechanical Analysis (CBA), consultants to this project; and Dr. Seif-Naraghi from Good Samaritan Medical Center, also a consultant to this project. Also in attendance was Dr. Richard M. Herman from Good Samaritan Medical Center, a physiatrist and senior researcher in the motor system, who has worked with MOCO on numerous previous human performance projects. The experimental studies for the five nonstandard conditions of limb movement were discussed at length. The major points were as follows. Consideration was given to whether the size of the subject should influence the weight of the load in the limb loading condition; no consensus was reached but the trend was toward using the same weight for all subjects. CBA presented a demonstration on a notebook computer of the Ariel Performance Analysis System (APAS) software that will be used in the muscle fatigue study, showing its operational capabilities on true biomechanical and electromyogram (EMG) data. For this condition, it was also decided that the subject's arm will have to be supported in, anticipation of the reach and a switch used to determine the initiation time, because the degrees of muscle fatigue expected will make it difficult for the subject to hold the arm in the air and impossible for the subject to hold the arm steady enough to obtain initiation time through software. Extensive discussion ensued pertaining to the justification of the muscle fatigue paradigm; clarification of potential problem areas was made by CBA. With regard to the perturbation condition, it was decided that the type of perturbation proposed was appropriate. One force and one time of application should be selected as the perturbation of interest and which must meet the design specifications. Perturbations with different forces and at different times will be interspersed as well as some highly unusual disturbance such as a loud noise and normal trials of standard reaching; these will serve only as random distracters and will not be analyzed as perturbations. It was observed that the subject may become habituated to the anticipation of a perturbation after a very short period of time, and that a degradation of the response may occur. This was deemed appropriate since in many functional activities, persons are in situations in which perturbations are occurring over a long period of time and the learned response is precisely what is of interest; caution was advised in the analysis of the results since there is an expected difference between the trials at the beginning and end of the session. It was further suggested that attention be paid to the interspersed normal trials since these may be quite different from those obtained in the baseline sessions and may contain significant information. No recommendations were made for the path obstacle condition. Comments similar to those made for limb perturbations were applied to the final condition of environment alterations. The target of interest will be the center target, with the alterations taking place to the right and the left; it may be possible to include a second target of interest, also with alterations taking place to the right and left, or to the right and then further to the right.
A large portion of the meeting was devoted to devising methods for establishing and confirming compatibility between the APAS and the Ascension Flock of Birds (FOB) for acquiring the kinematic data. The first issue was which points to digitize with the APAS in order to obtain data analogous to that obtained with the FOB. The second issue concerned the twist angles which traditionally are difficult to perturbation and the passive line tension have been measured. The goal is to minimize the passive force required to keep the line taut. Second, different motor alternatives for producing the perturbation have been considered. These are a stepper and a torque motor, with the goal of controlling the required tug without affecting the FOB magnetic tracking devices. Third, consideration has been given to where the disrupter will be located and how to route the line between the arm segment and the disrupter in a way that minimizes interference with the motor and the various wooden parts of the test station.
Since it is desired to apply the perturbation at a specific percent of the movement time, it is necessary to detect the initiation of movement in real time. Currently, the initiation of motion is calculated after the reach by analyzing the FOB data off-line. The limitation in processing of the FOB data in real time implies that a new means of determining the initiation of movement may be necessary. A number of different switches have been considered.
E. Visual Markers
The APAS is the kinematic recorder that will be used for the nonstandard condition of muscle fatigue, as discussed on P. C4, B. Objective 2, 1. Test Station. Although the APAS has been used in numerous other biomechanical studies, methods are being developed to improve the visual markers for the system to accommodate the special demands of this study. These markers must have the properties that: (1) they can be detected by an automatic computerized digitizing system and will reduce the amount of human interactive involvement with the process; (2) they will not interfere with the normal reaching task or with the performance of the intervening fatigue tasks; and (3) they will not move or be damaged due to the exercise or the perspiration generated. Some of the possibilities under consideration for markers on the hand, knuckle, and finger tip include the following. A latex, single finger surgical covering may be used to which either 3-M specialized reflective tape, some other reflective material, or reflective paint can be applied. Application of temporary or "press-on" nails to the finger nail in conjunction with different types and sizes of reflective balls are being considered. These options may be less effective because of the anticipated difficulty in executing normal movement during the reaching task and in holding the exercise bar during the fatiguing task. Efforts to modify a latex surgical glove and a bike glove were unsatisfactory and were rejected. Use of reflective paint or theatrical make-up applied directly to the skin are possible options. Preliminary use of a Dr. Scholl's Corn protector cushion on the knuckle is encouraging since the pad is flexible, and therefore easily conforms to skin and joint changes associated with flexion and extension of the hand during the reaching task and when gripping the exercise bar. A significant observation was that the corn pad did not move or slip when exposed to moisture, e.g., water; the next test will be exposure to exercise and the perspiration generated from that effort.
F. Comparison Study
A study design has been devised, as proposed on p. C6, C. Objective 3, c., d, for the comparison study of the trajectories obtained with the FOB and the APAS. The measurement errors or differences for the two systems could be caused by system set-up, resolution, and sensors and markers moving on skin. To isolate the possible different causes, a three-level testing and comparison method was designed.
The first level is a static comparison. Sensors and markers will be placed on rigid sticks fixed in space. The two systems will collect data simultaneously for several seconds, and the results will be compared.
(2) What kind of regression process should be used for fitting the function to the normalized paths? (Most likely, a weighted regression method based on the scatter area size will be developed.)
(3) Can a good fit to the experimental data be obtained for all targets using a function model with target position as parameters? (If so, the likelihood of predicting a path or path zone for an unknown target is high.)
(4) Why are the path scatter areas different for the different targets? (The optimization model may ultimately provide explanations.)
Three activities were conducted. These will be referenced to the Contract, Section C Statement of Work on pp. C2-C8, where relevant. These pages are also contained in the Contract Proposal.
A. Testing And Target Apparatus
A change in vendor has been made for construction of the empirical testing apparatus for the nonstandard conditions, as described on p. C3, A. Objective 1, 3. Path Obstacles and 4. Environmental Alterations; and of the target apparatus for the front panel of the leg workspace and for the muscular fatigue condition, as explained on p. C5, B. Objective 2,1. Work by Mr. Hal Layland, the vendor originally proposed, was not satisfactory. This aspect of the project has now been assumed by group that has had considerable experience in building original test items. All specifications and engineering diagrams have been forwarded to the current company. No additional expense will be incurred because of this change, and no delays are anticipated.
B. Comparison Study
Work on the comparison study of the trajectories obtained with the Ascension Flock of Birds (FOB) and the Ariel Performance Analysis System (APAS), as proposed on p. C6, C. Objective 3, focused on two tasks. A portable calibration device for the APAS system was built which can easily and inexpensively be-shipped between MOCO and Computerized Biomechanical Analysis (CBA), the consulting organization for this study. This device is used by the APAS to calibrate the three-dimensional area of interest, in this case, the immediate surroundings of the subject performing the reaching task. The device was required to be accurate and transportable. The use of PVC tubing, tennis balls and golf "wifffle" balls were shown to fit these necessary conditions. Metal and wooden structures were considered and rejected. In conjunction with the development of the calibration device, camera placements, lighting, and reflective markers were arranged for the test procedures. All filming specifications which could be considered and resolved in the CBA laboratory were evaluated in an effort to facilitate the experimental work which will be performed at MOCO for the actual study. The comparison study of the FOB and the APAS is now finalized. The comparison study of the trajectories obtained with the Ascension Flock of Birds (FOB) and the Ariel Performance Analysis System (APAS), as proposed on p. C6, C. Objective 3, is underway. The APAS equipment was delivered to MOCO, Inc.; this consisted of two cameras, the calibration cube and the various reflective markers. Communications with personnel from Computerized Biomechanical Analysis (CBA), the consulting organization for this study, determined the appropriate setup of the equipment in terms of the placement of the cameras, the use of lights, and the best camera settings. The equipment setup was further illustrated in a videotape prepared by CBA for MOCO which showed the same test session, including the calibration, subject-worn reflective markers, and subject reaches, through the eyes of each camera. Minor modifications were made to the MOCO FOB software, so that the FOB and APAS could be synchronized and used concurrently. Each trial proceeds as follows. The large light emitting diode (LED) display that is typically used to inform the subject of the number of the target to be touched for the impending trial is placed in the viewing field of both APAS cameras. The two cameras are started. The trial number is presented to the cameras by placing a clipboard containing the written number in front of the cameras for a few seconds. By means of a keystroke, three events occur simultaneously, namely, a tone is emitted, the LED display changes, and FOB data collection is initiated. The data are collected for six seconds. At this instance, the LED display changes again and the FOB stops automatically. The cameras are turned off manually. Thus, the two APAS frames in which the LED changes correspond to the initial and final data points of the FOB. This allows the signal points of one system to be matched in time with the signal points of the other system, and in addition synchronizes the two APAS cameras. A comprehensive three-day testing session was then conducted in which the FOB and the APAS recorded the same instance of each trial. A formal protocol was developed to permit comparison of the two systems. System performance was examined at four levels, each increasing the difficulty of obtaining system compatibility. The first level, termed static inanimate, was to determine the translational distances between two fixed points, fixed flexion angles formed by a goniometer, and fixed twist angle displacements made by a goniometer. In the second level, dynamic inanimate, trajectories were obtained for the three translation signals of randomly moving points, randomly changing flexion angles formed by two rods, and randomly changing twist angles produced by a rod. For the third level, static animate, a human subject held his arm in various fixed positions intermediate to a functional reach; the translational distances between selected body landmarks, joint flexion angles, and angles formed by body segments and the coordinate axes were obtained. At the fourth level, dynamic animate, the human subject made standard functional reaches; translation trajectories of selected body landmarks, joint flexion and twist angle trajectories, and trajectories of angles formed by body segments and the coordinate axes were acquired. Additional trials tested different variable shutter settings for the cameras. The protocol is given in detail in Appendix A.
A. Testing And Target Apparatus
Construction has been completed on the empirical testing apparatus for the nonstandard conditions described on p. C3, A. Objective 1, 3. Path Obstacles and 4. Environmental Alterations and on the target apparatus for the front panel of the leg workspace. Construction related to the muscular fatigue condition, as explained on p. C5, B. Objective 2, 1. Test Station, continued. The testing apparatus, namely, the disrupter mechanism, for the nonstandard condition described on p. C3, A. Objective 1, 2. Limb Perturbations, proved to be much more complicated than anticipated. The disrupter has gone through five iterations during the past seven months. A brief history and current status will be given.
The original design used a torque motor and a magnetic clutch arrangement. Modifications to the design included a roller clutch and a Neg'ator motor (constant force spring) for tensioning. Both the original and modified designs were discarded primarily because of the difficulty of maintaining proper tension on the subject's limb and the high cost of materials needed to implement the designs. This design phase was initiated in July, 1993 and continued to September, 1993.
The second design, initiated in October, 1993, used a combination of a linear actuator, solenoid, roller clutch, and Neg'ator motor. A prototype was completed in December, 1993 and was critiqued for the purposes of making improvements and corrections to this design. The finished prototype proved to be inadequate due to the limited amount of force that could be applied, the amount of arm movement required, and the actuation noise. The last limitation, noise, could have been corrected easily but after careful review, the other two limitations proved to be intractable.
A completely fresh approach was taken to the next design. Using the conceptualization of a spinning reel, this design would allow limb movement without any tensioning device while permitting control of the force and speed applied to the limit. The system involved an air driven rotary actuator and line gripping solenoid, plus an electrically driven motor to take up the slack in the supply reel. This design was critiqued in the first week of February, 1994 and was rejected because there would be too much line slack causing an unwanted jerk motion to be applied to the limb.
To correct the design deficiency of the spinning reel approach, it was decided to use a Neg'ator motor once again to tension the line attached to the limb. This then removed the jerk component of the force and also removed the need for a rotary actuator. The current design now consists of two linear, air driven solenoids to grip and pull on the line attached to the limb plus the Neg'ator motor to keep the line taut. The air driven approach was retained because such a pneumatic system can be controlled more easily with regard to speed and force.
The mechanical drawings of the current design are being completed. These drawings are necessary for the machining of the required lever arms, the specification of needed parts, and the final assembly specifications. In fact, the pneumatic devices, that is, the solenoids and regulators, have been ordered even before the completion of the drawings in the interest of saving time.
B. Comparison Study
Initial results of the comparison study of the trajectories obtained with the Ascension Flock of Birds (FOB) and the Ariel Performance Analysis System (APAS), as proposed on p. C6, C. Objective 3, were analyzed. In particular, selected trials from the formal testing session to examine static and dynamic inanimate and animate conditions were examined.
The first trial, from the static inanimate group, was to calculate the x, y, and z translational distances between two fixed points, with the distances pre-measured by a steel measuring tape. The following results (cm) were obtained.
x y z Tape 37.4 55.9 20.0 FOB 37.7 56.1 21.0 APAS 37.5 55.7 20.2
The second trial, also from the static inanimate group, was to calculate the fixed flexion angle formed by a goniometer. The following results (deg) were obtained.
FOB 135 APAS 135 angle 135
The third trial, again from the static inanimate group, was to calculate the fixed twist angle displacement made by a goniometer. The following results (deg) were obtained.
angle Goniometer 46 FOB 47 APAS 46
This result is particularly encouraging, since video motion analysis systems typically do not handle twist angles at all. The twist angle is obtained here through techniques specially developed for this application.
The fourth trial, from the dynamic inanimate group, was to obtain the trajectories for the x, y, and z translation signals of a randomly moving point. The results (cm) for the FOB and the APAS, respectively, are given in Figure la,b.
The fifth trial, again from the dynamic inanimate group, was to obtain the trajectories of a randomly changing flexion angle formed by two rods. The results (deg) for the FOB and the APAS, respectively, are given in Figure 2a,b.
A sixth trial, from the dynamic animate group, was to determine the complete kinematics of the arm during a standard functional reach made by a human subject. As a result of examining the trial, a method for processing this level was established. The FOB software developed by MOCO and the APAS software both perform a complete set of calculations
on the arm, consisting of translation trajectories of body landmarks, joint angle trajectories, and body segment angle trajectories. Although the two sets contain equivalent information, the trajectories in the two sets are different. Rather than rewriting the analysis software for either system, which would have been a major undertaking since both software systems are fairly complex, the following solution was adopted. The APAS data will be presented to MOCO as the x, y, and z translation trajectories of the end effector, wrist, elbow, shoulder, and sternoclavicular joint, and the two reflective markers on the forearm and the upper arm crosspieces, with no angle processing. MOCO will then use these to derive the trajectories that correspond to those used with the FOB system. In general, the preliminary view of the comparison data looks promising.
C. Comparison Study
Subject in Experimental Condition
A complete set of initial results of the comparison study of the trajectories obtained with the A Ascension Flock of Birds (FOB? and the Ariel Performance Analysis System (APAS), as proposed on p. C6, C. Objective 3, was analyzed with the APAS by Computerized Biomechanical Analysis (CBA), the consulting organization for this study. and was delivered to MOCO. In particular, examples from each level and sub-level of the formal testing session to examine static and dynamic inanimate and animate conditions was analyzed with the APAS. In all cases, the new method for processing was used. As described presented to MOCO as the x, y, and z translation trajectories of the reflective markers, end effector. or joints, whichever are appropriate.
Displacement of body's joints
The comparison study was then conducted at MOCO with six human subject in fulfillment of p. C6, C. Objective 3, b. and c. No attempt was made to control for age, gender, handedness, stature, or weight since the Objective was simply compare data simultaneously acquired by the FOB and the APAS. A session consisted of trials from Level III -- Static Animate and Level IV -- Dynamic Animate defined previously in Monthly Progress Report -- 8, using the software and task descriptions given in the text and the Preparation and Acquisition given in Appendix A of that report. Descriptions remain as given for Protocol R in the OWD Phase II Final Report for target panels; targets; positioning chair; system integration; placement of receivers and transmitter; procedures for anthropometric and environmental measurements; target definitions (all trials were performed in the arm workspace, on the front panel, with a normal speed, at the prescribed distance, using the right arm); basic task (except that for the Static Animate trials the limb was held fixed in an interior reach figuration); subject fixed point; subject positioning; typical trial (except that the subject was verbally informed of the target number, and for the Static Animate trials the limb was placed in an interior reach configuration prior to the computer tone and did not move); harness; body segment posture (except that for he Dynamic Animate trials, the subject was instructed to maintain the final configuration of the arm until the trial terminated, i.e., 6 seconds passed); and control, acquisition, and monitoring software (with the exceptions noted in Monthly Progress Report -- 8). Figures 3, 4, and 5, respectively, show the placement of the receivers and markers, the subject measurement sheet, and the protocol worksheet.
The testing required two laboratory technicians. The monitor prepared the subject and took the measurements, explained the test to the subject, announced the target number and type and the Level to the subject, presented the trial number to the APAS cameras, assured that the subject was performing as instructed according to all of the details of the protocol, and attended to the general comfort and safety of the subject. The instrumentation operator was responsible for calibrating the APAS, focusing the APAS cameras, running the FOB control program, starting the APAS cameras, starting the FOB data collection, verifying the incoming APAS video data on two television screens to which the cameras were connected, stopping the APAS cameras, verifying proper acquisition of the FOB results with the OWD analytical information and animation software, and manually recording trial repetitions and anomalies. Both technicians made whatever adjustments were necessary, prior to each trial, to assure that the FOB transmitter was located so that it could handle the four receivers throughout the motion, and that all the reflective markers could be seen by both APAS cameras throughout the motion. Each session took approximately 2 hours.
D. Comparison Study
Analysis of the initial results of the comparison study of the data obtained with the Ascension Flock of Birds (FOB) and the Ariel Performance Analysis System (APAS), as proposed on p. C6, C. Objective, was completed. The formal protocol for the initial study is contained in Appendix A of Monthly Progress Report -- 8, and explained in that report. As reported, the levels and sublevels of the protocol are as follows.
I. Static Inanimate A. Translation B. Flexion Angle C. Twist Angle II. Dynamic Inanimate A. Translation B. Flexion Angle C. Twist Angle III. Static Animate IV. Dynamic Animate
One example from each level or sublevel of the protocol was analyzed, for a total of eight examples.
In all cases, a new method for processing, justified in Monthly Progress Report -- 9, was used. This consists of the APAS data presented to MOCO as the x, y, and z translation values or trajectories of the reflective markers, end effector, or joints, whichever are appropriate, and MOCO using these to derive the values or trajectories that correspond to those used with the FOB system. The results will be presented next for the eight examples. Some of these were included in Monthly Progress Report -- 9, but will be repeated here using the new processing method.
The first example, from Level I.A., was to calculate the x, y, and z translational distances between two fixed points, with the distances pre-measured by a steel measuring tape. The following results (cm) were obtained.
x y z Tape 37.4 55.9 20.0 FOB 37.7 56.1 21.0 APAS 37.5 55.6 20.2
The second example, from Level I.B., was to calculate the fixed flexion angle formed by a goniometer. To introduce more of a challenge, the goniometer was perpendicular to the ground, but at a 4-5 degree angle outward with respect to the APAS cameras. The following results (deg) were obtained.
angle Goniometer 135 FOB 135 APAS 135
The third example, from Level I.C., was to calculate the fixed twist angle displacement made by a goniometer. Again, to challenge the systems, the goniometer was skewed with respect to the APAS cameras. The following results (deg) were obtained.
angle Goniometer 46 FOB 47 APAS 46
The fourth example, from Level II.A., was to obtain the trajectories for the x, y, and z translation signals of a randomly moving point. The results (cm) for the FOB and the APAS, respectively, are given in Figure 4a,b.
The fifth example, from Level II.B., was to obtain the trajectory of a randomly changing flexion angle formed by two rods. The results (deg) for the FOB and the APAS, respectively, are given in Figure Sa,b.
The sixth example, from Level II.C., was to obtain the trajectory of a randomly changing twist angle of a rod. The results (deg) for the FOB and the APAS, respectively, are given in Figure 6a,b. The seventh example, from Level III, consisted of a human subject holding his arm at the initial, interior, and final positions of a functional reach. The x, y, and z translational distances between each adjacent pair of FOB receivers was calculated. Wrist flexion/extension, wrist radial/ulnar deviation, and elbow flexion/extension angles were also calculated. Finally, the upper arm X-axis angle, upper arm Y-axis angle, and upper arm Z-axis angle were calculated, these being defined relative to an inertial moving coordinate system, with origin that is attached to and moves with the shoulder at the acromion and axes that remain directed as and parallel to those of the FOB transmitter. The following results (cm for translations and deg for angles) were obtained.