karinaagliullova

Today was an inspiring day that led me to progress further in my work. While I feel lonely working on the project by myself because my mentor, Mythra, went to India, I still can say that I am proud of myself. One big part of the task was accomplished, and I did it by myself without anyone's help. I finished the extraction of the x and y values from the centroid from the videos. I wrote some loops in Python that iterate through the bounding box coordinates while slicing the list that I created. It was a big step in progressing only my project because the next step requires me to be able to interpolate the frames. Even though I still have very little understanding of what that means and how to do it, I am more confident in my ability to figure that out and accomplish this task as well. It is because today's achievement boosted my confidence and reassured me of my growth in the technical side of computer vision. I am becoming more experienced as a researcher, and I cannot hold bac...

Today I learned a lot about the side of AI that I had never thought of. It is the AI's weaknesses and failures. As I am progressing on my project of extracting the centroid of every single video of my hand movements and interpolating the videos to 20 frames, I started noticing the way YOLO works. Overall, the idea of YOLO seems to be great, and it is the leading object detection neural network that exists. However, there are too many issues with it, and it will take many years, in my opinion, to develop a better version of it. The first thing that I would say is that YOLO is bad at detecting objects unless I follow very strict rules for recording the videos. It runs out that in all of my videos, my hand was too close to the camera. Basically, it was too close for YOLO to recognize it, so the confidence level was pretty low. As well as this, YOLO tried to detect other objects in the frame, such as a couch or a drawing on the whiteboard, or even my face as a hand. Because of these is...

I got more clarification on my current status with the project today. It was very helpful to have a meeting with Mythra, the person with whom I worked on the project. He explained to me that even though he is leaving for three weeks to go to another country, he will still be available via Zoom. However, there is another thing that worries me about this situation, which is the almost 10-hour time difference between the countries. He assured me that the communication would still be on the same level, though, so I feel better. This means that I will continue working on the Arms project instead of switching to the other project that I was considering, which is quality assurance in ultrasound. That would have been very inconvenient to switch to a completely different project in the middle of my internship because, from what I know, they use other tools and concepts to work on the ultrasound images. I feel lucky for being in such a situation and still keeping the same project. Additionally, ...

Today something unexpected happened. In the morning, all the undergraduate students had a meeting with Dr. Voyles. While I was reporting on the status of my project, I was told by Dr. Voyles that my supervisor and a mentor who is guiding me with my project are leaving to go to India because of personal issues. I was surprised as I did not expect that at all. I started doubting that I would be able to accomplish the goals for my project if I just worked by myself. I started feeling anxiety all over my body as my future in that internship was unclear. However, Dr. Voyles assured me that Mythra, my supervisor, is still going to keep in touch with me over Slack. Still, I had some worries because I am new to the field of AI and Robotics, so I started doubting myself about being experienced and knowledgeable enough to keep working on the project. I shared these doubts and worries with Dr. Voyles, and he understood me. He said that another person who works on the Arms project but a different ...

 As I am progressing on my work with hand gesture recognition, I now have my own dataset. My next step is to extract the x and y coordinates from the frames in the video of my hand going up, down, left, and right. How I understand this assignment is from the perspective of a bounding box that captures the hand palm in each frame, recognizing it, and then finding the center in that bounding box, which has x and y coordinates. It is also possible to find the height and width of the captured object, but as of now, I don't need it for my project. By extracting the x and y coordinates, I will be able to teach a robot to recognize my palm using the YOLO algorithm and move the robot's hands similarly to my hand movements. This task reminds me of teaching a child how to use their hands to hold a spoon, for example. Growing up with my niece, I was the one who would always take responsibility for teaching my niece various movements because it was exciting to me to be able to be in the ro...

 I had a fruitful day at work today. I was able to accomplish the task that was given to me, which was creating my own dataset of hand gestures. Throughout this process, I learned a lot about reinforcement learning and its use in AI. I learned how the robot recognizes specific objects by using YOLO's bounding boxes to identify the objects and learn the procedure of interacting with that object. I absolutely love to be exposed to this side of AI and be able to witness the beginning of robots learning and teaching various tasks. I cannot even imagine how far the AI field will go in fifty years. Even now, the robots can already learn the task just by watching a video, identifying an object, and repeating after that object. And after mastering the skill, they can start teaching humans that skill. I am amazed by the progress in AI. The more time I spend working in the Robotics Lab, the more I realize that I want to pursue a Master's degree in AI, robotics, or machine learning. This ...

Creating my own dataset turned out to be a hustle for me. I did not expect to struggle with it so much. The task itself sounded pretty simple and straightforward. The task was to record different types of hand movements using deep learning techniques. However, the obstacles were meeting me on each step of that task. At first, I had issues with various installations. There were a lot of packages, libraries, dependencies, and upgrades that I had to install, and every single one of those installations didn't go smoothly. I can still say that I am proud of myself because I eventually figured it all out. I had to use catkin to create my own workspace virtual environment, and it helped me understand ROS more. Eventually, I got to record my own videos, and the code, written with OpenCV and Yolo, worked just fine. I learned how to read, write, and show various video types. However, I faced another problem after recording the video. I am now not sure what to do next because the videos that ...

I am happy to announce that my work on creating a dataset is going great. Today I was able to start working with ROS, as well as learn such concepts as nodes, topics, master, messages, rosout, and roscore. I noticed that when I first started learning those concepts, I could not understand them fully, and it made me feel bad as if I was not doing good enough since it's just basic concepts of the ROS system. However, now I can see what a real struggle for me was. I realized that I didn't understand many concepts in theory before implementing them. I now can say that I am more of a practical person because I only understand the concepts once I apply them to real-world solutions. The same thing happened today. As I was reading the documentation for ROS, I couldn't figure out what all those terms are and how they are connected. But after I started implementing them directly into my project, I was able to understand that nodes are nothing but executables that communicate with eac...

It was an inspiring day for me today. It is because I started working on creating my own dataset of visual images. After studying visual images and video manipulation with Ros, YOLO, and open CV, I finally got a chance to apply the concepts and algorithms that I had been studying for almost three weeks in this internship. I am now in the process of recording my own dataset that includes videos of me displaying my hand moving up, down, left, and right. Every single move should contain twenty images that need to be detected by the program called RVIZ, which is a part of Ros. Today I can say that I am proud of myself as I no longer fear trying new things and not being afraid of failures because it is through the failures that I acquire perfection after every single try. I learned how to use a 3D camera to detect my own hand with the program, and it can send signals to the robot I am working with. It is fascinating to see the intelligence of the machines that can simply take the input and ...

Today in the Robotics Laboratory, I was exposed to learning a lot of new concepts and materials. Mythra came today to check on my progress, and he was happy to see me being done with the previous tasks and getting ready to start the new ones. He told me a lot of new information, ways to learn and practice the concepts, and my overall goal for this project. I was overwhelmed with all the information at first, and I even found myself a little frustrated at the end of his further explanations of the project. However, this time I did not let the feeling of frustration overcome the joy of learning new material. Throughout this internship, I am learning about myself more and more, and I can see the slow and steady progress in overcoming negative feelings, procrastination, and getting excited about the opportunity to work in one of the best universities in the country. Today I can say that I am proud of myself and my growth, as I did not dive deep into any negative feelings, but instead, I us...

Today was one of the hardest days I have had so far. The thing is, I got frustrated over a problem while working on it by myself. What I needed to do was to go over the installation of PyTorch and start with creating my own dataset. At first, I was full of enthusiasm to learn something new and practice my coding skills. However, by the end of the day, I could not help but get frustrated, and it was all because of the difficulties that occurred on my way. As I have just started working on Linux, I consider myself a beginner in using it and do not have much experience with it. Because of my little experience, I had difficulties later on. While trying to install many and many packages and libraries, I had a lot of errors, and some of them I could not solve for almost two hours. After that, I had difficulties executing any code that I wrote because of many other packages that were uninstalled. As well as this, I did not know where to begin in creating my own dataset as I was not given any ...

It was a day of inspiration and enthusiasm for me today. After listening to Dr. Voyles performing his presentation for undergraduate students, some interesting thoughts came to my mind. I was truly amazed by the intelligence and creativity of the people who work in the Robotics lab. Dr. Voyles was presenting all the projects that people in the lab are currently working on, and there are much more than what Mythra introduced us to. Some projects like "The Plane," "Mothership," "The Drone," and "Sustainable Precision Animal Agriculture" each deserve an entire lecture that could take hours, in my opinion. For example, "Mothership" is a tread/limb hybrid robot that uses treads as the primary mode of locomotion, augmented by active limb-like joints that articulate the discrete tread modules. This project's purpose is the application of Urban Search and Rescue. When I saw Mothership live, performing its motions, I could not believe that t...

I am happy to announce that I have finally started working on the Arms, specifically on the coding part. I have been waiting for it for a long time. Until today, I was assigned many readings and resources that I needed to get to know better, which was still a great way to learn a lot of valuable and meaningful theories and concepts. Throughout that period, I learned about computer vision, image processing, YOLO, basic image operations, object detection, a little bit of PyTorch, and different kinds of learning used for robots' learning. Although such a theoretical approach enhanced my knowledge and led me to a deeper level of understanding of all the concepts, I still did not have a chance to get involved in the actual process of the development of AI, which bothered me. It felt like I was worried that I was not learning enough because all I did was listen to lectures, read research articles, and watch tutorials. However, now I am delighted to share the news of getting into practica...