In order to optimize our linkage sizes to obtain our desired motions, we used the Phase Space motion capture system to record the relative angles of shooting motion.
Two L.E.D. sensors were placed around the shoulder, elbow and wrist joint in order to find the intersection where the joint would be. The data was processed using Excel and MatLab. A polynomial fit equation was derived based on the data and was fed through an optimization program on SolidWorks. Currently in progress.
For our project we used the PhaseSpace motion capture system setup in the ReNeu robotics research lab. We chose to incorporate motion capture data in our project to help us generate the desired movement in the exoskeleton. The motion capture data gave us the joint angles we wished to match with the exoskeleton. Then we were able to optimize the link lengths of the exoskeleton to better match the motion capture joint angles.
We took two different sets of motion capture data. The first set of data was used to gather the desired joint angles we wanted the exoskeleton to match. The first set of data consisted of Nhat performing the motion (throwing a basketball) three times. The second set of data consisted of Nhat wearing the exoskeleton, turning it on, and letting it run for three basketball tosses and returns. We then processed that information to plot the joint angles and compare.
For each set of data, we used 8 cameras pointing towards a centered spot where Nhat stood. Six were facing Nhat’s right, while the rest were facing his left. We used five LED in total to calculate the movement of the joint angles we were interested in. One was placed at the shoulder, this was used for ground. One was placed on the bicep by the elbow and another was placed on the forearm by the elbow, these two were used to calculate the elbow angle throughout the movement. One was placed on the forearm by the wrist and the last one was placed on the hand by the wrist, these two were used to calculate the wrist angle throughout the movement.
To calibrate the Phase Space system, we followed the calibration instructions provided by Phase Space. This included facing all cameras towards the center, giving each camera data points, and setting the origin. To collect data from the PhaseSpace system we used code provided by Priyanshu Agarwal. This piece of C++ code used Linux POSIX thread library to connect to the PhaseSpace computer and save each marker into a text file at the same time. This allowed the marker data to stay depended on each other, and not by shifted by time. The data that was saved into a text file was processed though Matlab to plot the joint angles of the elbow and wrist with relation to time.
There were several problems with the motion capture data. For the second experiment, one of the LED trackers, used to calculate the wrist joint angle, was covered by another link on the exoskeleton for most of the motion. This caused the data to be invalid for that section of the motion. Another problem was non-consistent x and y axis when we calibrated the PhaseSpace system. This caused the joint angles to be shifted by 90° or 180°.
Figure 1: Two L.E.D. markers are placed at each joint to estimate the true joint location.
Figure 2: Using motion capture to record the positions when shooting a basketball.