We made some progress on the robot arm. One of the grad students sat down with me and traced the errors back. After debugging for a little while, we realized that the value that shows the distance between two adjacent servos was wrong because the arm had been redesigned. These distance measurements were controlled by parameters, but we couldn't find where the parameters were stored. In order to find it, we tried tracing a parameter listed in the launch file. The parameter in question was called /path_svg_file. This parameter was being loaded from a file called RobotArmPath.svg. An svg file is a Scalable Vector Graphics file. When we opened the svg file we found this:
We were unsure as to what this meant. Our assumption was that this swirl showed the target for the robot arm meaning that its intended motion would be in a swirl.
Since this parameter didn't give us much luck we went back to the launch file. In it was a line that read:
<rosparam command="load" file="$(find me439robotarm)/src/robot_arm_info.yaml" />
This meant that the parameters that we were looking for were in a file called robot_arm_info.yaml. Upon finding this file, we found a list of all the parameters that we needed. From this file, I realized that the robotic arm was not intended to move in a swirl, it was supposed to hold a marker and draw that shape on a piece of paper.
The first thing we needed to fix was the distance parameter between servo 0 and servo 1. But, after I changed a parameter value in the yaml file, the value didn't change when I ran the program.
We also tested the Variable Stiffness Foot (VSF) on a subject this week. It was very exciting. We used the xsens suit (which is similar to motion capture but doesn't require cameras), motion capture, and the pacer robot. I got to drive the pacer robot in front of our subject which was very fun. We initially did tests in the lab with the motion capture cameras and force plates. After a couple of trials, we changed the stiffness of the foot from level 1 (very soft), to level 2 (slightly stiff), to level 3 (very stiff). We had hoped to change the stiffness while the subject was walking, but a wiper broke on the foot so we had to change the stiffness before each test. After we did some tests with the motion capture, we moved to the hallway and followed a path that had ramps and stairs. For each stiffness level, we followed the path with the pacer robot and without the pacer robot. The subject said that they liked the level 1 stiffness a lot for most of the tasks. It was very interesting to see some real testing in person.
We had a lab meeting on Thursday where all the grad students that went to the conference talked about what they took away from it. It was very interesting to hear about all the talks that they went to. Apparently, one scientist won the award for do it yourself biomechanics for building his own plane, flying it, and tranquilizing lions, cheetahs, and zebras in order to get tracking data on predator-prey interactions. Also, some animals can move with no brain activity. Their spine will become a Central Pattern Generator (CPG) and tell your body to do various repetitive motions.
I talked with Dr. Adamczyk about the robotic arm. As the summer is coming to a close, the top priority for the interns is WobbleBot and TorsoBot, not the robotic arm. Since there hasn't been a lot of progress with the ROS implementation, Dr. Adamczyk suggested that rather than continue working on that aspect, I should stress test the acrylic material on the redesign of the robot arm by moving it around constantly for an hour or so. This was important because we wanted to redesign all the robot arms to be like this one. If the material was not good, we needed to know before we redesigned 30 robotic arms with it. I put all of my notes on the ROS software onto a google drive folder and shared it with him for his robotics class.
First, I wrote a program that moved the robot in a repetitive cycle until I interrupted it so that I wouldn't have to move it myself for hours. This worked well until one of the pieces broke. Surprisingly, it was not the acrylic that broke, but a simple unscrewing of a piece on the robot's hand. Here is a video of it:
In order to fix this problem, we will be using lock-tight on the screws. The acrylic did not break for the entirety of the stress test. This meant that the material was good for the other robotic arms. Next week, we will be redesigning the other robotic arms.
Comentários