Both sides previous revisionPrevious revisionNext revision | Previous revision |
ires_journal5 [2018/07/08 18:11] – yuhanghe | ires_journal5 [2018/07/08 19:17] (current) – [New Korean] yuhanghe |
---|
==== New Korean ==== | ==== New Korean ==== |
| |
| This week, our Lab Manager, Dongbin, visited Leonard and me at KAIST. Dongbin introduced us to one of his underclassmen, who is currently working at Aerospace Engineering Lab at KAIST. He is currently working with the design of missiles. Dongbin and him shared many of their old story and memories with us. |
| |
\\ | |
| |
==== Korean Cultural Insight ==== | ==== Korean Cultural Insight ==== |
==== What I Learned about Myself ==== | ==== What I Learned about Myself ==== |
| |
I missed the rain. The constant overcast sky, spontaneous downpour, and occasional thunderstorm this week reminded me how much I missed the rain. The region in China where I was born and lived for 12 years of my live have similar weather conditions as South Korea. It rarely rained in Las Vegas, therefore, I did not have the opportunity to reflect on this point before. However, now I am in South Korea, where the summer experience a raining season similar to my birthplace. I finally realized how much I missed rain. | I learned that one of the biggest obstacle for my research process is my own stubbornness and inability to ask for help. One of the motto that I tried to guide myself is "work smarter not harder." However, I found myself often struggling with a problem and refuse to ask others for help. My mindset becomes entrapped inside a loop that keeps repeating "I can do it" to myself. More often than not, I will later find out that the problem I am struggling with have a known solution and there are others who could have helped me. I hope I can change this bad habit of mine. |
==== Project Status ==== | ==== Project Status ==== |
{{ :yuhang:ires2018:mocap1.jpg?500 |}} | {{ :yuhang:ires2018:matlabcode.png?500 |}} |
Last week, I learned how to use Vicon motion capture system from Hubo Lab member. At the start of this week, I sat up and calibrated 4 cameras for the mocap system and used them to capture myself performing some simple motions. I created 6 distinctive markers to track the position of my right shoulder, elbow, wrist, hip, knee, and foot. | Last week, I captured myself performing some simple motions using Hubo Lab's Vicon motion capture system. Therefore, this week I focused my effort on trying to convert mocap data into Hubo 2 joint data. |
{{ :yuhang:ires2018:mocap2.jpg?500 |}} | |
I recorded myself performing some simple motions like lifting arm, lifting leg, and extending arm. I learned how to use Vicon mocap system SDK to export the data points. | |
\\ | \\ |
| I decided to write my program in Matlab first because it is easier to test and visualize the result of my program. Matlab allows me to use linear algebra and visualize my results in plot and graphs, which would require third party library to implement in C++. |
\\ | \\ |
{{ :yuhang:ires2018:screenshot_from_2018-07-02_10-28-18.png?700 |}} | {{ :yuhang:ires2018:filtereddata.png?500 |}} |
The next step would be to use inverse kinematic and kinematic mapping to transfer these motions into robot joint angles. With the help of Hubo Lab members and Hyun Lee, I was able to find some examples of inverse kinematic functions with Hubo 2. I am currently studying these functions and learning how to use them to accomplish my goal. | The first step was to eliminate the difference between length of human arm and Hubo arm. I reduced the upper arm and forearm vectors in mocap data to unit vectors and multiply these vectors by the dimensions of Hubo arm. After plotting the data set, I noticed the noise and discontinuity in the data. I used filters to create a continuous slope and reduce the noise. |
\\ | \\ |
\\ | \\ |
{{ :yuhang:ires2018:screenshot_from_2018-07-02_10-33-55.png?700 |}} | Currently, I am working with the inverse kinematic function to transfer the filtered mocap data into joint position data for Hubo 2. I am still in the process of modifying the IK function to accomplish my specific need. I am also considering filter the joint data instead of vector data. |
Additionally, one Hubo Lab member asked me to collaborate with him on writing a tutorial on their wiki page for the process from installing Ubuntu OS and Xenomai to installing Choreonoid Simulator and writing a new PODO ALProgram. I will also upload the information to our wiki page once I am done with the process. | |