Motion Capture Study (Final Comparison)
In order to quantifiably determine the accuracy of our results, motion capture data was again recorded with Nhat attached to the final prototype. The idea was to show that the linkage synthesis results gathered from the initial motion capture data would closely match the results of the kinematics of the final prototype.
Below is the data collected by the motion capture system both before and after the implementation of the exoskeleton. Figure 1 shows the kinematics of Nhat's arm shooting a basketball. Figure 2 shows the kinematics of Nhat's arm as it is attached to the final prototype. For both figures, blue is the shoulder, green is the elbow, and red is the wrist. By comparing the joint angles for both systems, we can see how well the exoskeleton matched the motion we wanted.
Figure 1. Linkage angles captured from motion capture in initial study
Figure 2. Linkage angles captured from motion capture with exoskeleton
As can be seen, the shoulder angles of the two graphs do not match. This is acceptable, however, since we started each motion at arbitrary points. What is important to notice is the trend of the data. The ranges of motion for Nhat's shoulder have similar patterns. Both start at a given angle and then decrease in a parabolic fashion to a low-point, and then increase in the same shape until the shoulder returns to its initial angle. There are differences, however. In figure 1, we see that the change in angle of the shoulder is approximately 125 degrees. In figure 2, the angle change is only 50 degrees. Based on this data, it appears that the exoskeleton does not adequately capture the full range of motion of Nhat's natural basketball shot. In order to verify these results, however, we would need to collect additional data. It is also possible that the motion capture data is flawed, as anecdotally the range of motion seemed to be similar.
The elbow joint angles appear to match very well. For both figures, the elbows begin and end at approximately 100 degrees. In addition, both maximum angles for the elbow's range of motion peaks at roughly 175 degrees. Thus, based upon the elbow's data, the exoskeleton appears to copy Nhat's original kinematics fairly accurately.
As can be seen in figure 2, the motion capture data is limited for the wrist angles. Due to the design of the arm exoskeleton, the linkages around the wrist were obscured at various times throughout the range of motion of the basketball shot. As a result, no matter where we placed the sensors, only limited data could be retrieved. Nonetheless, the partial data gathered in figure 2 suggests that the motion was similar to the initial data. Perhaps the most noteable difference (as seen here) is due to the peak angle of the wrist. In the original data the wrist reached a maximum angle of approximately 60 degrees. However, in the final data collection, the wrist appears to achieve a maximum angle of almost 100 degrees. If this is the case, it suggests some errors in the synthesis. However, as with the shoulder data, further data collection is required in order to substantiate this claim.
Our results appear to be somewhat inconclusive. There are sections of the data that appear fairly accurate, whereas other portions seem to diverge significantly. There are two suggestions for future experimentation. First, a more rigorous data collection methodology is needed to ensure that the results can be trusted. If this is done and the results still appear to be inaccurate, then further synthesis will need to be run. In general it is highly possible that our synthesis code was not as robust as needed. Perhaps in the future we would want to include a larger range of possible link lengths and smaller step sizes in the iterative process. In the current code, these factors were kept small as the code took a substantial amount of time to run. Further effect could shorten the code's run time while also incorporating a more robust sample size.
In the end, however, the overall procedure appears to be promising.
Welcome to the University Wiki Service! Please use your IID (yourEID@eid.utexas.edu) when prompted for your email address during login or click here to enter your EID. If you are experiencing any issues loading content on pages, please try these steps to clear your browser cache.